Journal of Pathology Informatics Journal of Pathology Informatics
Contact us | Home | Login   |  Users Online: 486  Print this pageEmail this pageSmall font sizeDefault font sizeIncrease font size 




 
Table of Contents    
ABSTRACT
J Pathol Inform 2012,  3:37

Abstracts: Pathology Informatics 2012


Date of Web Publication09-Oct-2012

Correspondence Address:
Login to access the Email id

Source of Support: None, Conflict of Interest: None


Rights and PermissionsRights and Permissions

How to cite this article:
. Abstracts: Pathology Informatics 2012. J Pathol Inform 2012;3:37

How to cite this URL:
. Abstracts: Pathology Informatics 2012. J Pathol Inform [serial online] 2012 [cited 2017 Apr 23];3:37. Available from: http://www.jpathinformatics.org/text.asp?2012/3/1/37/102183


   Electronic Poster Session Top


Session-1

Wednesday, October 10, 2012 (3:20 - 4:10 pm)

Presented in the Regency Ballroom B,C,D

Biorepository Informatics and Management

Rocky Ackroyd 1 , Tim Marshall 2

1
Department of Pathology, Maine Medical Center/Spectrum Medical Group, MMC, Tissue Bank, Portland, Maine, 2Project Managment, NovoPath, Princeton, New Jersey.

E-mail: ackror@mmc.org

Content

Several efforts at the national level have attempted to develop standards and software for the management of biorepositories; however, there are few good commercially available systems that address all needs, including a comprehensive database infrastructure, a user interface that streamlines data entry, a simple search component for end-users, and integration into an existing, sophisticated anatomic pathology laboratory information system (APLIS).

Technology

NovoBioBank is a product of collaboration between Spectrum Diagnostic Services (Portland, Maine, USA) and NovoPath (Princeton, New Jersey, USA), a recognized anatomic pathology software system developer. A total of 26 years of collective experience, both in comprehensive tissue banking and software development, has led to an intuitive, robust, and information-centered system, targeted to fill the aforementioned void in the marketplace.

Design

The software will function as a stand-alone product for BioBank management and also be available as a module directly incorporated into the NovoPath pathology information system.

The development is occurring in three major phases:

Phase1: Comprehensive Database infrastructure development with search capabilities, inventory management, report generation, and sample tracking.

Status: Beta software Completed and being tested.

Phase2: Client-based order entry to include automation of sample selection, sample retrieval, labeling and shipping functionality, and tracking of orders with associated billing functionality.

Status: Currently in development.

Phase3: Plans to incorporate and/or reference clinical data in relationship to samples in the inventory. Linkage of whole slide images to Biobank samples.

Status: Future development.

Results

Beta software is currently being tested for both ongoing collections and retrospective biobanking programs (including paraffin archive material). Phase 1 project addresses the database infrastructure needs, search capabilities, and basic inventory management.

The software has the capabilities of tracking consent status, disease states, sample formats, availability status, immunohistochemistry, and molecular test results. Many quality-related data points, including sample weights, ischemic time, and pathology verification of samples (presence of normal or diseased tissue and tumor percentage when appropriate) are also integral to the software.

Conclusion

NovoBioBank is the first sophisticated and comprehensive informatics system specifically designed for biobanking that is integrated into an existing anatomic pathology laboratory information system (LIS) and seeks to fill the void in the existing commercial marketplace. The National Institute of Health and The National Cancer Institute have focused on this need in recent years, the result of which was the development of caBig and caTissue although they have seen limited adoption. NovoBioBank is software that is addressing this need and the development of this tool has been led by both a recognized Pathology Information System vendor and an established tissue bank operation.

Immunohistochemical Evaluation of Malignant Mesothelioma Tissue Microarray: National Mesothelioma Virtual Bank

Waqas Amin 1 , Malini Srinivasan 2 , Sang Yong Song 2 Nancy B. Whelan 1 , Anil V. Parwani 2 , Michael J. Becich 1

1
Department of Biomedical Informatics, University of Pittsburgh, 2 Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA.

E-mail: waa8@pitt.edu

Content

The National Mesothelioma Virtual Bank (NMVB) has proven to be a valuable tissue and data resource for the mesothelioma research community. The resource has provided hundreds of specimens, including fresh frozen, paraffin, and tissue microarrays (TMAs). We perform regular quality assurance on specimens that we shared with the investigators. We performed immunohistochemical evaluation of the University Pittsburgh Medical Center (UPMC) TMA with three key antibodies that are used in making the diagnosis of mesothelioma, and also compared the immunohistochemical assessment between manual scoring and image analysis.

Design and Technology

The UPMC TMA was assessed for the immunohistochemical expression of three markers used in the diagnosis of mesotheliomas: calretinin (N = 39), cytokeratin (CK) 5 / 6 (N = 33), and D2-40 (N = 37). Immunohistochemistry was evaluated by semi-quantitative (manual) scoring using light microscope (SS) and by automated image analysis (MS). For the semi-quantitative assessment, one experienced pathologist (SS) scored the immunohistochemical expressions for staining intensity and proportion of positive staining tumor cells was assessed. Staining intensity of tumor cells was classified as negative, weakly positive, moderately positive, and strongly positive. For image analysis, the TMA slide was digitized using Aperio ScanScope XTslide scanner (Aperio Technologies, Vista, CA, USA) at ×20 magnification. The tumor areas in each histospot on the TMA were manually annotated using Aperio's annotation software (ImageScope v11.1.2.760, Aperio Technologies). Calretinin and CK5 / 6 were assessed using Aperio's Positive Pixel Count Algorithm v9, and D2-40 was assessed using Aperio's Membrane algorithm v9.

Results







  • Calretinin staining was seen in both cytoplasmic and nuclear locations. CK5 / 6 stain was localized to the cytoplasm. D2-40 stain showed only membranous expression in our cases.
  • Based on the pathologist's scores, calretinin was positive in 31 of the 39 cases (80%), CK 5 / 6 in 15 of the 33 cases (46%), and D2-40 in 18 of the 37 cases (49%).
  • The positive agreement between manual scores and image analysis was 90% (35 / 39), 94% (31 / 33), and 95% (35 / 37) for calretinin, CK 5 / 6, and D2-40, respectively. There was substantial agreement between manual and automated scores for calretinin (kappa=0.614) and almost perfect agreement for CK5 / 6 (kappa=0.879) and D2-40 (kappa=0.892).


Conclusion

Our study confirms that the immunohistochemical staining pattern of mesothelioma in the NMVB UPMC TMA is similar to other studies. Our findings also show that automated image analysis provides similar results to manual scoring by pathologist, and provides a reproducible, objective, and accurate platform for immunohistochemical assessment of biomarker expression.

Current Workload and Storage Requirements Associated with Whole Slide Imaging

Victor Brodsky 1 , Jessica Pizzarello 2

1
Weill Cornell Medical College, New York, 2 State University of New York, Oneonta, NY. E-mail: victor.brodsky@med.cornell.edu

Content

Currently, only the institutions that chose to implement whole slide scanning are fully aware of the required time, effort, and cost. In this analysis, we survey the per glass slide workload and the scanner speed to enable organizations interested in establishing a scanning service to define the required number of full-time employee hours.

Technology

Aperio Scanscope AT (2012); Console version 101.0.4.413.

Design

One hundred and sixty-seven biorepositories and 579 breast consult slides were scanned at "40X" magnification (0.22 microns/pixel) and "×20" magnification (0.46 microns/pixel), respectively. The time spent in picking up the slides, loading them into the scanner, obtaining snapshots, adjusting the area to be scanned, modifying focal point locations, performing the scan, verifying image quality, rescanning slides, and typing in associated slide metadata was logged.

Results

At "×40," machine scan time took 22.96 min per slide. Rescanning of 2.45% of the slides identified during post-scan quality review resulted in 3.07 additional minutes of machine time per slide in the batch. All associated pre- and post-scan manual steps added up to 1.55 min per slide, resulting in the total time per slide of 27.58 min. The average file size of the "×40" images was 726 megabytes, with an average scanned area of 11,736,466,666 pixels, or 5.68 cm^2. Consequently, 4.86 min and 127.81 megabytes were spent per square centimeter of the scanned glass slide at "×40". At "×20," the average file size was 498 megabytes and the machine scan time took 3.44 min per slide, with 5.5% slides needing to be rescanned, thereby adding 0.19 min of machine time per slide. The manual steps bring the total scan time at "×20" to 5.18 min per slide.

Conclusions

A 7-h work day of one full-time employee followed by 16 h of machine scan time can result in a predicted maximum of 260 slides scanned at "×20" (likely much less due to human fatigue). At ×40, the machine scan time will extend to about 5 days. At this rate, a maximum of 13,000 to 67,000 slides can be scanned per year, using up 9-32 terabytes (at ×40 and ×20, respectively).

Effect of Pen Marking on Glass Slides for Whole Slide Image Scanning

Ryan A. Collins, Anil V. Parwani, Walid E. Khalbuss, Jon Duboy, Liron Pantanowitz

Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA. E-mail: collinsra2@upmc.edu

Content

Whole slide imaging (WSI) is the digitization of glass slides for archival, education, consultation, and diagnostic purposes. This digitization process often requires an autofocus step to accurately capture (scan) microscopic material on the glass slide, followed by a stitching procedure to correctly merge scanned fields into one confluent digital image. In the practice of anatomic pathology (e.g. cytotechnologists screening slides), marker pens are often used to make "dots" on the coverslip to indicate regions of interest. The aim of this study was to determine whether such pen marks (dots) interfere with the aforementioned steps in the imaging process.

Technology

Multiple WSI scanners-Aperio Scanscope XT (Aperio Inc, Vista, CA, USA), Hamamatsu Nanozoomer (Hamamatsu Corp., Bridgewater, NJ, USA), and Zeiss Mirax (Carl Zeiss Microimaging, Thornwood, NY, USA). WSI viewing software (Aperio ImageScope 10.2, Vista, CA, USA).

Design

Ten glass slides representing different pathology materials were selected including five surgical pathology cases (H and E stain) and five cytology cases (Pap and Diff-Quick stains). All slides had several (6-24) pen marks (dots) on the coverslip surface. Pen colors included blue, black, and green. Dotted slides were scanned on three different WSI scanners (Aperio, Hamamatsu, and Zeiss). Slides were then cleaned with ethanol to remove all dots and scanned again using the same machines. All slides were scanned at ×20. Digital WSI images were examined for aberrations (focus, stitching, and other artifacts), comparing the exact regions (using ImageScope coordinates) between dotted and cleaned slides.

Results

Pen marks on glass slides resulted in stitching errors, focus problems, and other artifacts (pixilation and scanner skipping large portions of the slide) to varying degrees when scanned with different WSI instruments [Table 1].

Conclusion

Digitization of glass slides by all tested WSI scanners was adversely affected by the presence of pen marks (dots) on the glass slide surface. The Nanozoomer had the least issues regarding disturbed image stitching and focusing secondary to dotting of glass slides. These data indicate that pen marks should be removed from glass slides prior to scanning. The results also show that imaging algorithms of WSI scanners display varying capabilities of dealing with foreign markings on the glass slide surface.



The Impact of "Field-of-Gaze" on Digital Pathology Reviews

Catherine M. Conway 1 , Neil O'Flaherty 2 , Avi Rosenberg 1 , Alina Nicolae 1 , Gregory Riedlinger 1 , Marios Gavrielides 2 , Stephen M. Hewitt 1

1
Laboratory of Pathology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, 2 The Division of Imaging and Mathematics, Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, US Food and Drug Administration, Silver Spring, Maryland. E-mail: catherine.conway@nih.gov

Content

With the advancements in whole slide imaging technology, the need for calibrated, high-resolution monitors has become a necessity in the evaluation of digital slides. There is a general misconception that the monitor and therefore the amount of tissue being displayed must be as large as possible. We have investigated the impact of the size of the field-of-view on a pathologists review. In addition, we have observed that the boundary color that surrounds the digital slide may impact immunohistochemical evaluation.

Technology

Digital slides were captured at 40× magnification using a Nanozoomer 2.0 HT (Hamamatsu, NJ, USA) instrument. Images were viewed using a web-based software system (SlidePath, Ireland). Slides were reviewed on a consumer grade, 27" LED backlit LCD monitor (Samsung), which was calibrated and color managed to international color consortium standards.

Design

Twenty-five whole sections were stained immunohistochemically for Her-2 protein expression utilizing the Herceptest™ (Dako, CA, USA). Four pathologists reviewed the whole sections on the 27" monitor and scored membrane staining intensity according to the package insert. After a 2-week wash-out period, the pathologists re-scored the 25 cases. However, rather than using the entire 27" monitor the area of the screen displaying the slide was greatly reduced, to be comparable to the field-of-view observed under a 20× objective. Images were also displayed with a variation in background color (black, grey, and white).

Results

We have observed that pathologists do not use the entire computer monitor when reviewing digital slides. In fact, they treat the digital slide as they would a glass slide under a microscope, by moving the field-of-interest into the lower center of the monitor which is in their field-of-gaze. Provisional results demonstrate that there is no substantial benefit in reviewing digital slides on a large monitor. In addition, we have found that variability in color surrounding the field-of-interest may increase intra-observer variability.

Conclusion

Pathologists do not need large monitors when performing digital reviews. Typically, the area under review at any one time should be comparable to the field-of-gaze observed under a microscope at 20×. However, consistency in color surroundings and the quality of the monitor will impact observer variability.

Visual Ergonomics and Complex Pathology Reports

Philip R. Foulis 1 , Carlos A. Muro-Cacho 1 ,

Leah B. Stickland-Marmol 1 , Matthew Banas 2 ,

Steven L. Luther 3 , Yao Djilan 4

Departments of 1 Pathology and Laboratory Medicine, and 2 Systems Redesign, James A Haley Veterans' Hospital, 3 Consortium for Health Informatics Research (CHIR) and the VA Center of Excellence: Maximizing Rehabilitation Outcomes, Tampa, 4 Office of Disability and Medical Assessment (DMA), St. Petersburg, FL. E-mail: philip.foulis@va.gov

Content

Human factors engineering involves the integration of computer science, behavioral science, patient safety, and visual ergonomics. It can be used to discover the most efficient way to design electronic pathology reports and to facilitate content extraction. The purpose of a Cancer Protocol is to convey diagnostic and prognostic information to health care providers. Implicit in this objective is the clinician's ability to efficiently and effectively extract critical information. To date, there is limited literature describing the optimal presentation of complex laboratory results.

Technology

Using an audience response system (Turning Technologies, Youngstown, Ohio, USA), we measured answer accuracy and response time to questions regarding information contained in the CAP Fallopian tube Cancer Protocol; purposefully selected because it is a complex report of a rare neoplasm and unfamiliar to most clinicians.

Design

Eight formats containing all required elements were created with variations in capitalization and justification. Four of the formats included a box containing selected diagnostic and prognostic information, which was duplicated in the text component of the protocol. A variety of individuals in the medical field, including students, cancer registrars, and oncologists, were randomly assigned one of the formats and queried about protocol content. We measured the speed of accurate responses to a series of questions to determine the best format.

Results

Respondents correctly answered questions more rapidly when using protocols with boxes, independent of other formatting (capitalization and justification). Furthermore, answers containing limited discrete data (e.g. 5, positive, pT3) are better discriminators of report design. No gender differences were observed.

Conclusions

This methodology, using measurable responses to a set of questions, identifies some favorable design elements to optimize efficiency and accuracy of data extraction. Even within a synoptic report, a summary box highlights critical parameters and provides easy access to information necessary for clinical decision making. In the design of future formats, concisely presented data should be favored. This is the first study that investigates speed and accuracy of data extraction from Cancer Protocols. Human factors engineering concepts should be considered when designing complex pathology reports.

Use of Desktop Sharing Software for Dynamic Telecytology Review of Pap Smears: Results of an International Effort Linking Peru and the United States

Nicholas C. Jones 1 , Erika Escalante 2 , Brenda J. Sweeney 1 , Barbara Winkler 3 , William Tench 4 , Rosemary Tambouret 1 , Nancy Joste 5 ,

Patricia Wasserman 6 , Ronald N. Arpin III 1 , Nasera Hassan 7 , David C. Wilbur 1

1
Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts, 2 Cervi-Cusco Clinic, Cusco, Peru, 3 Mount Kisco Medical Group, Mount Kisco, NY, 4 Palomar Medical Center, Escondido, CA, 5 University of New Mexico, Albuquerque, NM, 6 Columbia University Medical Center, New York, NY, 7 Duke University Medical Center, Durham, NC. E-mail: ncjones@parnters.org

Content

The Cervi-Cusco clinic in Cusco, Peru, has a large number of international volunteers who collect, process, and interpret Pap smears who come for 1 or several weeks at a time but only for 4-5 weeks per year. Although the clinic is now served by a local gynecologic cytology screener, there is no pathologist available for the majority of the time. Weekly web conferencing with desktop sharing was used to provide live images to volunteer pathologists and cytotechnologists in the USA, who interpreted cases identified as potentially abnormal by the cytology screener. Telecytology interpretations allow for more frequent and timely diagnosis and guide treatment for cervical disease in a population that would not otherwise have access to pathologist interpretations. The accuracy of interpretations rendered was measured by colposcopic and histological follow-up.

Technology

Digital camera attached to a microscope, computers with internet access and software to actively view camera images, and desktop sharing "webinar" program (GoToMeeting, Citrix, Santa Barbara, CA, USA).

Design

A mixture of conventional smears and SurePath TM cervical cytology specimens was interpreted by at least one pathologist using the web conference. Slide and field-of-view selection were based on the review of the CerviCusco cytology screener.

Results

Three hundred and ninety-three cases have been reviewed by this process from 6 / 10/2011 to 8 / 17/12. Twenty-five cases have had colposcopy and surgical biopsy follow-up, of which 24 cases show correlation, with one case being an under-interpreted as LSIL with a histological diagnosis of squamous cell carcinoma. In this case, despite the discrepancy, the abnormal result still led to the patients' colposcopy, biopsy, and correct treatment.

Conclusion

The results of the telecytology review are promising, showing that the real-time digital camera images transmitted over web conference are adequate for interpretation. The outcome of the weekly web conference is leading to improved patient care in Peru. In addition, this process has the benefit of providing ongoing educational feedback to the cytology screener in Peru. The study is limited by the lack of follow-up in the population deemed to be negative and the slow rate of obtaining histological follow-up in the population identified as abnormal.

Evaluation of Laboratory Information System Automated Systemized Nomenclature of Medicine Coding

Arivarasan Karunamurthy 1 , Anil V. Parwani 1 , Anthony Piccoli 2 , Liron Pantanowitz 1

1
Department of Pathology, 2 Clinical Department Systems - Information Services Division, University of Pittsburgh Medical Center, Pittsburgh, PA. E-mail: karunamurthya@upmc.edu

Content

Systematized Nomenclature of Medicine (SNOMED) is a widely used coding system with applications expanding beyond pathology. SNOMED coding has evolved into several versions since its release in the 1970s, along with changes in diagnostic pathology terminology. Accurate coding is a prerequisite in order to reap the benefits of SNOMED-coded information. Most anatomic pathology laboratories rely on their laboratory information system (LIS) to accurately generate SNOMED codes. The aim of this study was to evaluate the accuracy of automated SNOMED coding performed by our anatomic pathology LIS.

Technology

Automated coding was performed by the anatomic pathology LIS (CoPath Plus version 3.2, Cerner) autocoder application (MSM) using SNOMED II version. Manual coding was performed using the printed SNOMED II version (CAP, Illinois, published 1979).

Design

We selected 150 surgical pathology reports that were consecutively signed-out. For these cases, SNOMED codes generated by the LIS were compared with manual coding using the same SNOMED II version. Comparisons were restricted largely to topography (T), morphology (M), procedure (P), and diagnosis (D) axes. Discordant cases were collaboratively evaluated with the LIS vendor.

Results

There were 224 specimen parts with diagnoses for these 150 cases. In 165 parts (74%), automated and manual coding was concordant. For the 59 discordant cases, the reasons why codes did not match were attributed to absent SNOMED codes in 41 (69%) parts, incorrect codes in 14 (24%) parts, and irrelevant extra codes in 4 (7%) parts. No codes existed for relatively newer diagnostic terms such as thyroid papillary microcarcinoma and sessile serrated adenoma. Examples of incorrectly autocoded cases were due to different terminologies (e.g. Barrett's esophagus autocoded as gastric metaplasia and Helicobacter pylori coded as Campylobacter, NOS), words not used by the autocoder (e.g. benign, lateral, squamocolumnar), or incorrect punctuation (e.g. hyphens). Autocoding-modified text (i.e. punctuation corrected) improved autocoding results.

Conclusion

Autocoding of SNOMED codes by the LIS is intended to increase productivity and reduce coding errors. Autocoding can be optimized by adhering to diagnosis text entry rules (e.g. use of correct punctuation and spellchecker) and pathologist validation of autocoded results at sign-out. These data show that the autocoder correctly coded only three quarters of our current surgical pathology cases. The majority of improperly coded cases in our study could be ascribed to formatting that hindered the autocoder and/or possibly that an outdated SNOMED version was employed. Therefore, to keep up with newer diagnostic terminologies and classifications, accurate autocoding of anatomic pathology cases by the LIS should employ current versions of SNOMED (e.g. SNOMED CT).

Automated Mapping of Laboratory Elements to LOINC Using Previously Established CPT Code Mapping

Ellen King 1 , Neil Shah 2 , Adam Harris 2 ,Teresa Bosler 2 , Richard Scheuermann 2

1
University of Texas Health Science Center San Antonio, San Antonio, 2 University of Texas Southwestern Dallas, Dallas, TX. E-mail: kinge@uthscsa.edu

Content

The University of Texas Southwestern is composed of several different hospitals, each with its own individual laboratory. In addition, there are more than 25,000 unique laboratory elements that are both redundant and lack unity. An automated approach was designed to help unify these elements into a single data warehouse by assigning each laboratory element, a unique LOINC identifier (Logical Observation Identifiers Names and Codes).

Technology

An automated LOINC matching program was written using the Ruby programming language.

Design

The algorithm first clusters the laboratory elements according to name and CPT code. A match is created when the name of the laboratory element is at least 85% similar to a LOINC identifier using the Dice Algorithm. Next, the sensitivity of the algorithm is increased by matching data that would have otherwise been eliminated against a generated list of synonymous test names (e.g. "LDL" and "Cholesterol in LDL"). Specificity is increased by linking CPT codes to their respective LOINC identifiers using previously established mapping created by UMLS and Mayo Laboratories. Unmatched data are further screened against the complete LOINC lab panel. If still however, no match is found, the "top 2000" LOINC identifiers are then searched for a possible match. If more than one match is found per lab element, then regular expressions for fluid type, units, and time elements are used to filter the data. Finally, a true match will be defined if there are less than six LOINC matches per laboratory element.

Results

Of the original 25,377 laboratory elements, this program matched over 15,000 lab tests (62% of total) with nearly 80% accuracy.

Conclusion

The LOINC mapping program, created at our institution, has proven to be successful in matching LOINC identifiers to our laboratory elements. Ongoing refinement of the program will lead to improved results.

Evaluation of the International Academy of Cytology (IAC) Virtual Slide Library

Liron Pantanowitz 1 , Ritu Nayar 2 , Manon Auger 3 , Fernando Schmitt 4 , Patricia Wasserman 5, David C. Wilbur 6 , Walid E. Khalbuss 1

1
Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA, 2 Department of Pathology, Northwestern University, Chicago, IL, 3 McGill University, Montreal, Canada, 4 University of Porto, Porto, Portugal, 5 Acupath Laboratories, Plainview, New York, 6 Harvard University, Massachusetts General Hospital, Boston, MA. E-mail: pantanowitzl@upmc.edu

Content

In recent years, web-based learning has been enhanced by offering digitized slides online, since they are more interactive than static images. As a result, virtual atlases using whole slide images (WSI) are now offered by several pathology societies. Since October 2010, the International Academy of Cytology (IAC) began offering its members access to an interactive virtual slide set. The aim of this study was to evaluate the effectiveness of this virtual cytopathology atlas.

Technology

A ScanScope XT WSI scanner (Aperio, Vista, CA, USA) was employed. Images were stored on a 2TB Storage Area Network server. Hyperlinks to these images were incorporated on web pages created with Joomla 2.5.6 (New York, NY) and images made viewable using Aperio's ImageScope viewer. Web analytics were collected using Nova Quiz 2010.

Design

International authors were invited to contribute cases. Their glass slides were scanned and digitized slides shared online as a virtual slide library at http://www.cytology-iac.org/educational-resources/virtual-slide-library. For each case, the WSI and representative static images were offered with interactive questions and answers. In a subset (n = 11) of these cases, annotation ("dotted regions of interest") on the WSI was also provided.

Results

A total of 42 cases were posted including Pap tests and non-gynecological cytology. [Table 1] compares the number of viewers' hits on static images and WSI, as well as the proportion of correct diagnoses they entered for these cases. Correct diagnoses for the annotated cases were on average 38% per case for WSI "without dots" compared to 64% per case for WSI "with dots" and 60% per case for static images.



Conclusions

Virtual slide libraries are an attractive educational resource pathology societies can offer online. They promote web-based learning and boost web traffic to their website. These data show that viewers preferentially viewed static images. This may be related to user preference, reluctance to use virtual slide technology, and/or possibly limited Internet connectivity. Cytology WSI with annotation performs better than those without annotation likely because they require only interpretation skills from the participants, rather than both screening and interpretive skills for WSI without annotation. Digital images are clearly a great boon to pathology education and their role in e-learning is anticipated to grow.

"Unusual" Phlebotomies: Does Collection Above an IV or Saline Lock Result in Variance in Commonly Tested Analytes?

Heidi L. Paulin 1 , Christopher Naugler 2

1
Department of Pathology, Dalhousie University, Hailfax, Nova Scotia, 2 University of Calgary and Calgary Laboratory Services, Calgary, Alberta, Canada. E-mail: heidi.paulin@dal.ca

Content

Traditionally, "unusual" phlebotomies (UPs), such as those drawn above a peripheral IV or saline lock, have been relatively contraindicated due to presumed suboptimal performance. This has been challenged by several studies, but most were small or utilized specific patient populations. Using a large quality assurance database in a broad patient population, we aimed to evaluate the difference between common blood analytes drawn via a UP (above an IV or saline lock) as compared to those drawn by conventional phlebotomy utilizing the patient as their own control.

Technology

Calgary Laboratory Services is the sole provider of laboratory services in the Calgary, AB, Canada metropolitan area, with a catchment of approximately 1.4 million persons and an annual reportable output of approximately 23,000,000 test results. Characteristics about all UPs performed by our staff between December 1, 2009 and September 30, 2011 were recorded in our Millennium laboratory information system (Cerner Corporation, North Kansas City, MO, USA). Statistics were computed using SPSS version 19.0 for Windows (IBM, Armonk, NY, USA) setting α at 0.05.

Design

We queried our quality assurance database for results of six common blood analytes in patients who had undergone a UP and compared them against results from the same patient taken by conventional phlebotomy within 7 days. Results were analyzed by paired t-test and univariate logistical analysis looking for statistically and clinically significant differences between the mean test results and UP type.



Results

We studied a total of 7380 paired IV phlebotomies and 4330 paired saline lock phlebotomies. The mean differences between results of UPs versus conventional phlebotomies were statistically significant for most analytes, but none of the differences were large enough to be clinically meaningful. The coefficient of variation for each analysis was small, indicating that the majority of the samples were close to the mean difference [Table 1]. Conditional logistical regression controlling for age and sex showed variable statistical significance for each analysis, but the mean differences were clinically insignificant without an identifiable pattern.

Conclusion

We demonstrate that selected routine blood analytes show no clinically significant differences when drawn by UP as compared to conventional phlebotomy.

Failure Mode and Effects Analysis for the Surgical Pathology Laboratory

Luigi K. F. Rao, Thomas M. Gudewicz

Department of Pathology, Massachusetts General Hospital, Boston, MA. E-mail: luigirao@gmail.com

Content

The surgical pathology section is an essential, intricate division within virtually all laboratories involved in delivering anatomic pathology services.

It encompasses several different areas from when the sample is obtained in vivo from the patient through the microscopic evaluation by a pathologist. From patient identification and association to accessioning to gross examination to tissue processing, numerous steps are required before patient tissue can be represented on a glass slide and a diagnosis can be rendered. Inherent to this multistep nature are the potential for errors at any level of the entire process. A variety of guiding principles that have been used within business models have been applied to similar stepwise processes including arguably the most well repudiated philosophy in lean manufacturing, best known perhaps in the setting of car maker Toyota's push for elimination of waste. Failure mode and effects analysis (FMEA) also has been utilized by industry to aid in decision making, goal prioritizing, and risk avoidance. FMEA was initially developed in the 1940s by the United States military and later adopted by the National Aeronautics and Space Administration in the 1960s. Further advancement for FMEA was done with the civil aviation sector, followed by helicopter and eventually automobile manufacturing. This gradual and diverse adoption suggests that industry application of FMEA is not limited but may be modified to fit a particular entity's operations and workflow improvement programs.

Technology

QA Assistant (tm) software, Microsoft Excel program.

Design

Twenty-seven functional areas with distinct physical locations and/or separate tasks were initially found encompassing the anatomic pathology laboratory. FMEA was specifically applied to the biopsy service grossing workflow, with risk priority numbers (RPNs) being assigned.

Results

RPNs determined varied widely with high values associated with sentinel event type instances (e.g. 600 for patient misidentification due to wrong container label) to moderate-to-low values related to delayed turnaround times (e.g. 350 for a case more appropriately grossed in a different section and 60 for additionally sampled tissue given a new accession number).

Conclusion

FMEA's potential applicability for the surgical pathology laboratory's multilayered, time- and labor-intensive processes are immense. From effectively minimizing catastrophic failures and reducing probability of significant errors by determining severity level, occurrence frequency, and detection reliability, FMEA can offer to those who working in surgical pathology a major and effective tool in improving quality metrics. By enhancing patient safety, lowering errors, and reducing wastes and associated costs, FMEA should be regarded as a valuable, structured tool within the cache of all who play an integral role in evaluating and promoting excellence within the anatomic pathology laboratory and its many systemic processes.

Retrieving Molecular and FISH Data from Pathology Reports: A Comparison of Synoptic and Natural Language Search Tools in a Commercial Anatomic Pathology Laboratory Information System

Geoffrey H. Smith, Shelley Caltharp, Deborah F. Saxe, Karen Mann, Charles E. Hill, Alexis B. Carter

Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA. E-mail: geoffrey.smith@emory.edu

Content

Molecular and fluorescence in situ hybridization (FISH) laboratories may use anatomic pathology laboratory information systems (APLIS) to report results on specimens also being morphologically examined. Some laboratory information systems (LIS) search tools operate on structured information (e.g. discrete synoptic data), whereas others operate on unstructured information (e.g. free text). In this study, we compared the quality and ease of data retrieval between the synoptic search (SS) tool and natural language search (NLS) tool of our LIS.

Technology

Electronic synoptic reports were built in our LIS (Cerner CoPathPlus v3.3, Kansas City, MO) for molecular and FISH tests performed on anatomic pathology specimens.

Design

Results were recorded in both interpretation (free text) and result (discrete synoptic data) fields. SS and NLS queries were implemented for: (a) BCR-ABL1 translocation by polymerase chain reaction (PCR), (b) BCR-ABL1 translocation by FISH, (c) clonal T-cell receptor gamma chain gene rearrangement by PCR, and (d) two or more abnormal results by the myeloma FISH panel. The SS result set was validated by comparison with manual record review over a 1-month period. Subsequently, the SS and NLS result sets were compared over a 12-month period for (a), (b), and (c) and a 5-month period for (d). NLS criteria were iteratively modified to maximize intersecting cases, except for (d).



Results

During validation, SS returned 142 of 143 cases selected by manual review. The missing case did not have a synoptic in error. During comparison of SS and NLS results sets (n = 1928), concordant cases (n = 1891) were considered to be accurate. Discrepant cases (n = 37) were manually reviewed and characterized as shown in [Table 1]. After multiple iterations of NLS criteria definition, the analytical performance of both search tools was similar for (a), (b), and (c).

Conclusion

Analytical performance of SS and NLS in our LIS was similar. However, multiple iterations of tedious NLS criteria definition were required to achieve this level of performance. The SS tool required one iteration of criteria definition, and the criteria were subjectively far less complex. Therefore, searches (a), (b), and (c) were easier to implement with the SS tool. Query (d), involving a logical grouping operation, was difficult to implement with both tools.


   Scientific Sessions Top


Session-1

Clinical Subtrack

Scientific Imaging & Applications Development

Wednesday, October 10, 2012 (7:30 am - 8:55 am)


Regency Ballroom A

Subclassifying the Spectrum of Abnormal Endometrial Proliferation with Secretory Change: Supervised Automated Image Analysis as a Tool for Quantitative Assessment of Ki67 Index

Grzegorz T. Gurda, Alexander S. Baras, Robert J. Kurman, Toby C. Cornish

Department of Pathology, The Johns Hopkins Hospital, Baltimore, MD.

E-mail: ggurda1@jhmi.edu

Content

Abnormal endometrial proliferation exists in a spectrum from focal crowding to hyperplasia to carcinoma. Diagnostic morphology, however, can be obscured by secretory changes. Here, we assess if Ki67 index correlates with morphologic diagnoses and how well human estimates and automated image analyses subclassify the entities within this spectrum versus the gold standard of time-intensive manual counts.

Technology

Cell profiler (Broad Institute); Image Scope (Aperio Technologies).

Design

Seventy cases of endometrium with secretory changes were collected: baseline secretory endometrium, focal crowding, hyperplasia (simple/complex) and endometrial carcinoma. Corresponding Ki67 immunostains were photographed with three representative ×100 fields per case. All images were analyzed with automated software and 1 per case counted manually or by averaged 'eyeball' estimate by three independent pathologists. Statistical analyses were carried out with Graph Pad Prism.

Results

The Ki67 index closely parallels the step-wise progression of abnormal endometrial proliferation with secretory change; similar rates are obtained by manual counting, estimates, and automated image analyses [Table 1]. Overall, there was a high concordance between manual counts versus estimate and automated analysis (93%, 100%). At intermediate Ki67 rates (10-75%), automated analysis may be more accurate (100% vs. 86%), likely due to human overestimate. Human eye may be superior at identifying secondary structures (glands, vessels, etc.), whereas automated software is more accurate at numerical counts. Time allotment per case: 10 min (manual), 30/s min (automated) and 15-20 s (estimate).



Conclusion

Automated analysis is a fast and useful tool in assessment of Ki67 index, particularly at boundaries of overlapping, but differentially managed diagnostic entities. Combination of human-automated methodology (supervised automated counting), may further improve its utility. Future directions: Assess inter versus intra specimen variability as a diagnostic marker; crowdsourcing (Amazon.com mTurk platform) as an ancillary tool in supervised automated counting.

Automated Whole Slide Image Screening for Prostatic Adenocarcinoma by Use of Spatially Invariant Vector Quantization

Jennifer Hipp, Lakshmi P. Kunju, Ulysses J. Balis

Department of Pathology, University of Michigan, Ann Arbor, MI. E-mail: jenhipp@gmail.com

Content

Spatially invariant vector quantization (SIVQ) has exhibited broad utility in the detection of subtle architectural and nuclear features, making a compelling case for the exploration of its utility to screen (identify) prostate cancer in core biopsies. Given the time-consuming task of screening, automated detection would be of significant utility to the diagnosing pathologist in transforming workflow from a time-intensive screening task to a directed-review/confirmatory activity.

Technology

Digital whole slide images were obtained from six prostate core biopsies with prostatic adenocarcinoma (Gleason score 3 + 3 = 6). Use of a region-of-interest extraction tool, dCore, followed by use of an image aggregation tool, ImageMicroArray Maker, allowed for the generation of a montage of diverse examples of prostatic adenocarcinoma, atrophy, adenosis and benign glands, and finally, high grade prostatic intraepithelial neoplasia (PIN).

Design

Since the morphologic criteria for prostate cancer includes both architectural features and the presence of cytologic atypia, image analysis was performed at low and high magnification (×8 and ×40, respectively). Two vectors in tandem were utilized to identify malignant epithelial architectural features. Ring vectors specific for cytologic atypia were chosen to identify nuclear enlargement, hyperchromasia, and prominent nucleoli on the high magnification field views.

Results

SIVQ-based ring vectors were able to efficiently identify the cytologic features of cancer (35/37 malignant glands with a sensitivity and specificity of 87% and 94%, respectively), as well as high-grade PIN (owing to both diagnostic entities having the same cytologic features). This same constellation of ring vectors was successful avoiding false-positive detection of areas of adenosis, atrophy, or benign epithelium. Architecturally tuned vectors exhibited a high rate of detection of most small infiltrating glands (122/179 malignant glands) but also with a small false discovery rate for a minority of admixed atrophic glands (7/38 F.P. for benign glands).

Conclusion

SIVQ-based detection is able to efficiently identify the architectural and cytologic features of prostate cancer with a higher sensitivity for nuclear atypia. We attribute the lower sensitivity of architectural vectors detection to the variation in size and color of cancerous glands, and to the difficulty of distinguishing malignant from atrophic glands at low magnification. This study provides pilot data that use of the SIVQ algorithm as a stand-alone solution holds promise for the development of automated whole slide image (WSI) detection.

Development of a Computer-Aided Decision Support Tool for Diagnosis of Micropapillary Urothelial Carcinoma

Steven C. Smith 1 , Jason Hipp 1 , Jerome Cheng 1 , James Monaco 2 , Anant Madabhushi 2 , Jesse McKenney 3 , Lakshmi P. Kunju 1 , Ulysses J. Balis 1

1
Department of Pathology, University of Michigan, Ann Arbor, MI, 2 Rutgers University, Piscataway, MJ, 3 Stanford University, Stanford, CA.

E-mail: smithscs@umich.edu

Content

Micropapillary urothelial carcinoma (MPUC) is a distinctive, aggressive variant of urothelial carcinoma (UC). Recent studies have examined the prognostic value of micropapillary morphology in UC; however, the results have exhibited significant interobserver variability in both diagnosis and threshold for amount of MPUC versus non-MPUC morphology that is associated with poorer prognosis. Thus, we investigated the feasibility of a computer-aided approach to assist pathologists in the diagnosis and quantitation of this entity.

Technology

Spatially invariant vector quantization (SIVQ) is an image analysis algorithm that uses circular textural vectors for pattern recognition in digital slides. Image microarrays are tiled digital images of constrained size and resolution containing histopathologic features of interest, prepared from digital whole slide images.

Design

We iteratively developed vectors targeting two recently studied diagnostic features of MPUC: (1) retraction artifact, reportedly 100% sensitive, and (2) multiple nests within the same lacuna, reportedly 95% specific. We then applied the algorithm, using these vectors, to image microarrays constructed from quadruplicate representative MPUC and non-MPUC areas of digital whole slide images of 10 MPUC cases culled from a recent consensus report on MPUC morphology.

Results

For detection of retraction artifact, the algorithm identified classic MPUC morphology, showing 12.0/5.6 [95%CI]-fold greater detection in MPUC as compared to non-MPUC regions. For the multiple nests within the same lacuna feature, the algorithm was more specific for MPUC, showing 18.8/8.9-fold greater detection in areas of MPUC.

Conclusion

Our findings suggest that this novel strategy may provide a useful decision support tool to pathologists in the diagnostically challenging setting of MPUC. The automated feature identification process is tailored to recapitulate the thought process of the pathologist, based on matching sensitive or specific features of MPUC. We are currently testing the utility of combining vectors for both features to optimize identification and quantitation of MPUC morphology. Future studies will test the utility of this approach for the standardization of interobserver reproducibility for diagnosis and quantitation of MPUC.

Session-2

Discovery Subtrack

Scientific Imaging & Applications Development

Wednesday, October 10, 2012 (7:30 am - 8:55 am)


Acapulco Room

MIGIS: High Performance Spatial Query System for Analytical Pathology Imaging

Ablimit Aji 1 , Qiaoling Liu 1 , Fusheng Wang 2,3 , Tahsin Kurç 2,3 , Joel Saltz 2,3

1
Department of Mathematics and Computer Science, Emory University, Atlanta, GA, 2 Department of Biomedical Informatics, Emory University, Atlanta, GA, 3 Center for Comprehensive Informatics, Emory University, Atlanta, GA.

E-mail: aaji@mathcs@emory.edu

Content

Systematic analysis of large-scale microscopy images can generate tremendous amount of spatially derived quantifications for micro-anatomic objects such as nuclei and blood vessels. Most queries on the results are spatially oriented, highly computational, and data intensive. For example, there are queries on spatial cross-matching of multiple sets of segmented objects and on spatial proximity between micro-anatomic objects and local neighbor density. We report our recent progress on developing a highly scalable and efficient query system for supporting large-scale analytical queries on pathology images.

Technology

We are developing a MapReduce-based framework MIGIS to support cost-effective high performance spatial queries on commodity clusters. The data are staged on Hadoop distributed file system (HDFS) which supports reliable storage with replication. We are building a spatial query engine by extending CGAL computational geometry library and spatial index library. We will take advantage of the open source MapReduce framework Hadoop for parallelizing and scheduling query execution.

Design

MIGIS consists of a spatial query engine RESQUE which runs spatial query by building spatial indexes on the fly. Data files with segmented boundaries and extracted spatial features are merged and staged onto HDFS. Spatial queries are expressed as "map" and "reduce" functions and parallel execution is performed by Hadoop. Effective job optimization is performed to minimize query execution bottleneck due to data skew on MapReduce.

Results

MIGIS supports following queries: spatial cross-matching query for algorithm evaluation, nearest neighbor queries to find closest blood vessel for each cell, and high-density queries to find regions with high concentration of nuclei, such as pseudopalisades. We tested our experimental study with datasets of 18 and 50 whole slide images, respectively. Our study demonstrates high scalability of the queries on MapReduce.

Conclusion

The big data from medical imaging-the vast amount of spatially derived information generated from pathology image analysis-shares similar requirements for high performance and scalability with enterprise data, but in a unique way. Our system MIGIS demonstrates an effective and highly scalable solution that can run on cost-effective commodity clusters. This can be used for algorithm sensitivity study and whole slide image analysis.

Evaluation of 655 Patent Applications Related to Digital Pathology

Ioan C. Cucoranu 1 , Anil V. Parwani 1 , Ronald Weinstein 2 , Jonhan Ho 1 , Liron Pantanowitz 1

1
Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA, 2 University of Arizona, Tucson, AZ. E-mail: cucoranuic@upmc.edu

Content

Digital pathology is reshaping the practice of pathology. Development of applications for telepathology and image analysis will play a major role in the future adoption of digital pathology. There have been major technological advances in this field over the last two decades. Most of these developments have been captured in the form of patents, to protect the intellectual property of inventors. Our aim was to systematically evaluate patents related to digital pathology.

Technology

Patents published in databases at www.uspto.gov (United States Patent and Trademark Office, Alexandria, Virginia, USA), were searched using Google Patents (Google Inc., Mountain View, California, USA). Data were extracted and analyzed with Papers software (Mekentosj BV, Aalsmeer, Netherlands).

Design

The United States Patent and Trademark Office public database was queried using the following keywords: "Computer assisted diagnosis and pathology," "digital image analysis and pathology," "digital imaging and pathology," "digital imaging and pathology and data security," "digital microscopy," "digital pathology," "telemicroscopy," "telepathology," "telepathology and data security," "videomicroscopy," "virtual microscopy," "virtual pathology," "whole slide imaging," "whole slide scanner and pathology." Google search engine was used to facilitate the download of data into the Papers application.

Results

A total of 655 patent applications related to the field of digital pathology were identified, including 413 (63%) patents specific to pathology use, and 242 (37%) patents also usable beyond pathology. Of the 413 pathology specific filed patents, 297 (72%) were awarded and 63 of them expired (due to non-payment of maintenance fees or term end). There were also 52 abandoned applications, 63 pending patents, and 1 rejected application. Several (28) initial telepathology patents were related to television. The number of pathology-specific patent applications quadrupled in the last decade: 68 (1991-2000) versus 323 (2001-2010). The number of patents actually awarded almost tripled during these periods [Figure 1]: 66 versus 208.

Conclusion

This study shows that there have been many patent applications related to digital pathology in the last two decades. Of these, the number of patents specific to use in pathology almost quadrupled in the last decade. Most (2/3) of them are related to telepathology (including whole slide imaging, data management, networking, data security, and user interface), whereas the rest (1/3) are related to image analysis. Based on these observations, technological advances related to digital pathology, especially tools for image analysis are expected to increase in the future.



Early Experience with Optical Coherence Tomography: A Novel 3D Imaging Modality in Pathology

Jeffrey L. Fine, Ioan Cucoranu

Department of Pathology, University of Pittsburgh, Pittsburgh, PA.

E-mail: finejl@upmc.edu

Content

Optical coherence tomography (OCT) and related three-dimensional (3D) microscopy techniques can potentially allow pathologic diagnosis directly from tissue, without glass microscope slides. Our early efforts have focused on correlation with histopathology and optimizing the imaging process, first with an ophthalmic system and now with a newer pathology system. We present our experience with OCT imaging as we begin systematic studies with the technique.

Technology

Modified full-field OCT (Light CT, LLTech, Paris, France), using a motorized stage to capture and stitch together multiple ×10 fields of view into a single image or image stack. Its resolution is about 1 μm (transversely and axially). Resulting images are viewed using included software. H and E slides were scanned using a WSI system (Aperio XT, Vista, California, USA).

Design

Fixed and fresh tissue was scanned using several different imaging parameters. H and E slides from the imaged tissue were subsequently prepared and compared. In addition to evaluating images for diagnostic features, scans were evaluated for other features such as depth into tissue, scan time, and usefulness of quality parameters.

Results

Tissue samples from 22 patients (about 50 tissue samples) were scanned. Large-area images were limited to one (or few) focal planes into the tissue, but thicker 3D image stacks were possible using fewer fields of view. Resolution appears adequate to discern small features (fat septa in breast, recognizably mucinous glands in an ovarian tumor), but images appear to have limited contrast in many tissue types. Both fixed and fresh tissue worked well, though there appear to be subtle differences in image quality related to firmness of fixed tissue.

Conclusion

OCT is an exciting imaging modality and many histopathologic features are recognizable in our images. Further studies are needed to better establish its utility in clinical settings such as "room temperature" frozen section, or even in-vivo imaging. Image size, capture speed, and other technical issues exist but will likely be solved with hardware improvements if a market for clinical applications develops. Other medical specialties are already developing 3D microscopy applications, and if pathologists can become involved, we will be better able to adapt as future medical diagnostic work evolves.

A Method for Quantitative Analysis of Ki67 with Enlarged Follicles in Follicular Lymphoma

Bruce Levy 1 , Yukako Yagi 2 , Abner Louissaint Jr. 3

1
Department of Pathology, University of Illinois at Chicago, Chicago, IL, 2 Pathology Imaging and Communication Technology Lab, Massachusetts General Hospital, Boston, MA, 3 Department of Pathology, Massachusetts General Hospital, Boston, MA. E-mail: bplevy@uic.edu

Content

In many cases, the size of an area of interest or image analysis software may restrict simple quantitative analysis. For example, while analyzing cases of follicular lymphoma stained for Ki67, enlarged/coalesced follicles could not be analyzed from a single captured image. A method was developed to subdivide these follicles into smaller segments that were analyzed individually and then mathematically reconstituted to determine Ki67 positivity for the entire follicle.

Technology

A Hamamatsu Nanozoomer 2.0 HT scanner with associated scanning and viewing software, ImmunoRatio image analysis software (http://imtmicroscope.uta.fi/immunoratio/), and a basic desktop computer and monitor were used in this study.

Design

Slides of follicular lymphoma stained with Ki67 were scanned with a Hamamatsu Nanozoomer 2.0 HT scanner with NDP.scan software at ×20 objective magnification. A pathologist examined the slides using NDP.serve viewer and divided the follicles into two-dimensional grids of 0.1 mm 2 squares that were individually exported. Each square was analyzed for the percentage of Ki67-positive nuclei using ImmunoRatio image analysis software. The pathologist defined the boundaries and determined the area of the follicle in each exported image, and the software calculated the percentage of nuclei positive for Ki67. The average percent positivity for the entire follicle was then calculated utilizing the formula:

Average positivity = Σ(A n*P n)/Total area where A is the Area of follicle in each square, P, the percent positivity for each square, and n is the number of squares making up a follicle.

Results

Comparing the analysis of follicles analyzed in their entirety and then divided into between two and four segments validated the method. A total of 23 cases of follicular lymphoma were evaluated using this method. Areas analyzed per case ranged from 1.3519 mm 2 to 5.1404 mm 2 . The Ki67 positivity was calculated to be between 3.6% and 82.1%.

Conclusion

The grid method presented in this paper demonstrates a process in which larger areas that cannot be quantitatively analyzed in one image can be broken down and evaluated in a systematic method. It could be automated, providing a quantitative result for an entire slide quickly and reliably. The process is not limited to Ki67 analysis, but could be applied to many different quantitative analyses. It can also be used to compare different areas from the same case for variations in positivity, which was noted in this study, but not further analyzed at this time.

Session-3

Education

Wednesday, October 10, 2012 (7:30 am - 8:55 am)


Toronto Room

Revitalization of Small Group Pathology Teaching in a Medical School Curriculum Incorporating Informatics Tools

Philip J. Boyer, Dana M. Grzybicki, Robin L. Michaels, Robert L. Low

Department of Pathology, University of Colorado Denver, Aurora, CO. E-mail: philip.boyer@ucdenver.edu

Content

Pathology instructional content is a cornerstone of the organ system-focused years 1 and 2 curriculum of the University of Colorado School of Medicine and includes 20 two-hour "small group" sessions (156 medical students divided into 8-10 groups) with faculty and residents serving as preceptors. The small group sessions had been essentially unchanged for approximately ten years. There was lack of uniformity of content and instructional model of the sessions; materials were outdated; student scores in the pathology component of standardized examinations lagged behind scores in other disciplines; and evaluation of the small group sessions by students, highly variable among instructors, ranged from high praise to disappointment to virulent diatribes.

Technology

Distribution of the text- and image-based content of the curriculum incorporated Microsoft Word and PowerPoint documents, Adobe Acrobat Pro PDFs, standard HTML Web pages, and the Blackboard content management system. Specimens used for macroscopic instruction were acquired from surgical pathology and autopsy cases to supplement the extensive but outdated teaching collection.

Design

Case-based educational modules were developed for each of the 20 small group sessions, incorporating didactic content and gross specimens to compliment and supplement material covered in lectures, guided by the National Board of Medical Examiners (NBME), United States Medical Licensing Examination (USMLE) topic lists. Outcomes assessment included (1) tracking of student performance on pathology content of a customized version of the NBME comprehensive basic science exam (CBSE) taken after completion of the first year of medical school and on Step 1 of the USMLE taken after the second year and (2) evaluation of feedback from students for each "block" of the curriculum and on graduation questionnaire (GQ) surveys.

Results

There has been an improvement in pathology component scores on the CBSE and NBME Step 1 examinations; generally strongly positive comments about small group sessions and practice questions on course evaluation; and reduction in complaints about the pathology course on student GQ surveys.

Conclusions

Work to date has met the primary objectives of improving and standardizing content. Advances during the 2012-2013 academic year will include incorporation of required Blackboard-based "pre-test" quizzes for each session.

Active Learning for Pathology Image Analysis

Lee A. D. Cooper 1 , Christina Appin 2 , Rami Yacoub 3 , David A. Gutman 4 , Hyun Ju Choi 4 , Jun Kong 1 , Fusheng Wang 4 , Carlos S. Moreno 2 , Robin Bostick 3 , Daniel J. Brat 2 , Joel H. Saltz 4

1
Center for Comprehensive Informatics, 2 Department of Pathology and Laboratory Medicine, 3Department of Epidemiology, Rollins School of Public Health, 4 Department of Biomedical Informatics, Emory University, Atlanta, GA.

E-mail: lee.cooper@emory.edu

Content

Object classification is a fundamental problem in pathology image analysis. We have developed an active learning system to perform object classification tasks in whole-slide imaging applications. This system enables users to iteratively refine classification rules through a series of computer-guided experiments. The classifier selects examples for user review, and feedback is collected to refine the classifier at each iteration. This machine-driven approach forces users to label ambiguous examples and also hides algorithm complexity from end users, enabling scientific collaborators to be engaged more easily.

Technology

Sections of glioblastoma tissue were stained with Hoescht and quantum dot immunohistochemistry to identify mTOR, phospho-Rb, and Ki67. Slides were digitized at ×20 objective magnification on a 3D Histech panoramic scanner. Nuclei were segmented using gradient flow tracking, and each cell/nucleus was represented with 73 features describing morphology and protein expression. LogitBoost was used to classify cells and provide classification confidence scores. Scores are used to select samples for review to create a uniformly distributed set of examples from obvious to ambiguous.

Design

Examples are presented in groups of fifty, evenly split between classes, and classification and segmentation errors are recorded (random class assignment initially). Review continues until enough examples from each class have been identified to train the classifier. Two experiments were performed: (1) Identification of cells with any single positive signal and (2) identification of cells with co-expression. Classifiers were trained until the pathologist was satisfied and then validated on 500 randomly selected cells.

Results

The first experiment achieved 99% accuracy in discriminating single positive cells. This classifier was trained in four iterations reviewing 300 total cells. Of these, 76 required correction, with 73 corrected in the first iteration. The classifier for co-expression did not converge during the initial iteration due to a lack of co-expressing examples. A total of 600 cells were reviewed in this iteration.

Conclusion

The active learning framework is a promising approach to object classification problems, but requires further investigation to determine its limitations. The initial training iteration is problematic for applications with rarely encountered objects since examples are randomly chosen initially. Future studies will investigate improving sampling methods and more complex applications.

Preparation and Training for Interinstitutional Whole Slide Image Clinical Consultation:

A Process Analysis


Nicholas C. Jones, Elena F. Brachtel, Rosemary H. Tambouret, Chin-Lee Wu, Eugene J. Mark, G. Petur Nielsen, Rosalynn M. Nazarian, Lyn M. Duncan, Gregory Y. Lauwers, David C. Wilbur

Department of Pathology, Massachusetts General Hospital and Harvard Medical School, Boston, MA. E-mail: ncjones@partners.org

Content

Prior to beginning an interinstitutional whole slide image (WSI) subspecialty consultation pilot, a number of volunteer pathologists and one WSI technician needed to gain experience with scanning and interpreting WSIs and emulating the clinical workflow with a digital pathology platform. This organic process of training new pathologists across multiple subspecialties was continuously evaluated.

Technology

The Corista Digital Pathology Platform (Corista LLC, Concord, MA, USA, web browser-based WSI viewer and case management software), Mirax MIDI WSI Scanner, networked PCs.

Design

An imaging technician scanned and entered cases from the normal clinical workflow of volunteer subspecialist pathologists into the digital platform. The pathologists were asked to interpret the cases on WSI before their normal glass slide interpretations. No washout period was used in order to emphasize potential differences between digital and glass slide interpretations. Case selection was not random; a variety of smaller cases were selected in order to minimize the impact on the pathologist's time. WSI and traditional interpretations were compared, pathologists were surveyed as to their thoughts on the process and technology, and technical data were gathered.

Results

One hundred and seventy specimens, spread among nine pathologists, were reviewed . One hundred and sixty-three (95.9%) of the WSI interpretations correlated with the glass slide interpretations. Six (3.5%) had minor discrepancies (no patient care impact), and one (0.6%) case had a major discrepancy (potential patient care impact). Pathologist responses to the training methodology and platform were highly positive, with the time involved in WSI interpretation being cited as a concern, primarily due to computer speed, network speed, and input device limitations.

Conclusion

Though the goal was not to validate WSI use, the correlations between digital and glass interpretations were excellent. The individuals involved were comfortable with the technology and primed for WSI telepathology consultation. As computer hardware and network bandwidth increase at predictable rates and pathologist WSI screening speed increases with experience, it is expected that these limitations will decrease with time.

The History of Pathology Informatics: A Global Perspective

Seung Park 1 , Anil Parwani 1 , Raymond D. Aller 2 , Lech Banach 3 , Michael J. Becich 4 , Stephan Borkenfel 5 , Alexis B. Carter 6 , Bruce A. Friedman 7 , Marcial Garcia Rojo 8 , Andrew Georgiou 9 , Gian Kayser 10 , Klaus Kayser 11 , Michael Legg 12 , Christopher Naugler 13 , Takashi Sawai 14 , Hal Weiner 15 , Dennis Winsten 16 , Liron Pantanowitz 1

1
Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA, 2 Department of Pathology, University of Southern California, Keck School of Medicine, Los Angeles, CA, 3 Department of Anatomical Pathology, Walter Sisulu University, Mthatha, South Africa, 4 Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, PA, 5 International Academy of Telepathology, Heidelberg, Germany, 6 Department of Pathology, Emory University, Atlanta, GA, 7 Department of Pathology, University of Michigan Medical School, Ann Arbor, MI, 8 Department of Pathology, Hospital General de Ciudad Real, Ciudad Real, Spain, 9 University of New South Wales, Centre for Health Systems and Safety Research, Australian Institute of Health Innovation, New South Wales, Australia, 10 University Freiburg, Institute of Pathology, Freiburg, Germany, 11 Charite, Institute of Pathology, Berlin, Germany, 12 University of Wollongong, Centre for Health Informatics and e-Health Research, New South Wales, Australia, 13 Department of Pathology, University of Calgary, Calgary, Ontario, Canada, 14 Department of Pathology, Iwate Medical University School of Medicine, Iwate Perfecture, Japan, 15 Weiner Consulting Services, Florence, OR, 16 Dennis Winsten & Associates, Inc., Tucson, AZ. E-mail: parks3@upmc.edu

Content
"If you don't know where you came from how can you know where you're going?" The history of pathology informatics is a tale with many dimensions. At first glance, it is the familiar story of individuals solving problems that arise in their clinical practice of medicine. Under the surface, however, lie powerful forces-technical, regulatory, societal, and beyond-that have all played their part in molding our discipline into what it is today. The aim of this project was to solicit a historical account of the evolution of pathology informatics from informaticists around the world.

Technology

E-mail correspondence; Microsoft Word (Redmond, WA, USA).

Design

Leading figures in pathology informatics from all over the world (Africa, the Americas, Asia, Australia, Europe) were requested to submit historical material (articles, monographs, personal remembrances, photographs) regarding local progress made in pathology informatics.

Results

A review of the submitted materials over a 1-year period indicates the following trends: (1) Early efforts in pathology informatics stemmed from the USA and Europe; (2) pathology informatics is now a global discipline, with almost all continents driving developments in the field; (3) the advent of the Internet was a disruptive event with significant impact on pathology informatics; (4) the prevalence of different technologies in various regions correlates with intangible factors (e.g., regulatory concerns) and tangible factors (e.g., operational costs).

Conclusions

This project resulted in the first global compilation of the history of pathology informatics. It clearly demonstrates that intimate knowledge of our past is a critical foundation for wise stewardship of our future. This historical document is intended to serve as a baseline for future developments and refinements in the field. Further work is required to capture missing details to complete this treatise.

Session-4

Clinical Imaging

Wednesday, October 10, 2012 (7:30 am - 8:55 am)

New Orleans Room

Report from the 2 nd International Scanner Contest 2012

Peter Hufnagl 1,2 , Norman Zerbe 1,2 , Alexander Alekseychuk 3 , Frederick Klauschen 1 , Gian Kayser 4 , Marcial Garcia Rojo 5 , Arvydas Laurinavicius 6 , Yukako Yagi 7 , Thomas Schrader 8

1
Institute for Pathology, Charité - Universitδtsmedizin Berlin, 2 University of Applied Science Berlin, 3 Computer Vision and Remote Sensing, TU Berlin, Berlin, 4 Institute of Pathology, University of Freiburg, Freiburg, Germany, 5 Servicio de Anatomia Patologica, Hospital General de Ciudad Real, Ciudad Real, Spain, 6 National Centre of Pathology, Vilnius, Lithuania, 7 Massachusetts General Hospital, Boston, MA, 8 Department Informatics and Media, University of Applied Sciences Brandenburg, Brandenburg, Germany. E-mail: peter.hufnagl@charite.de

Content

The International Scanner Contest (ISC) understands its mission to promote digital pathology and virtual microscopy in research, education, and diagnostics.

Technology

Several algorithms and applications have been developed to measure and compare focus and image quality, speed, color fidelity, and geometric distortions.

Design

The 2 nd ISC 2012 contained five domains to evaluate different capabilities of participating systems. Each vendor was free to choose which of the available domains to attend:

High Throughput

Sets of 35 slides of different origin randomly mixed had to be scanned automatically under daily-use conditions. Preparation and scanning time have been measured and a focus-corrected speed was calculated.

Quality

The same 10 slides had to be scanned by each vendor in a limited time with best quality. Blinded quality evaluation was done by experienced pathologists applying a 9-tired scoring system.

Fluorescence

Two artificially stained monochrome slides (quantum dots) had to be scanned in a limited amount of time. Contrast, contrast-noise ratio, signal-noise ratio, and sharpness were measured to compare results.

Image Analysis

Twenty spots located on a Ki67 stained breast cancer TMA had to be scanned automatically. Subsequently, participants had to quantify the amount Ki67-positive tumor cells in selected regions. Results were compared to manual quantification by a reference panel of pathologists.

Technical

Two artificial slides, an IT 8.7/1 color target and a calibrated micrometer raster, had to be scanned. Resulting virtual slides were analyzed for color fidelity and geometric distortions.

Results

In the contest, we evaluated seven scanning systems (Metafer-VSlide-SFx80, NanoZoomer-HT-2.0, Pannoramic-Desk, Pannoramic-250, TISSUEScope-4000, UltraFastScanner-UFS, VS120-S5) from six vendors (3DHistech, Hamamatsu, Huron, Metasystems, Olympus, Philips) participating in 32 tests that were all passed successfully.

Conclusion

The ISC 2012 has a great benefit for the digital pathology community, because it offers a standardized and comprehensive means of evaluation, and therefore, a way for manufacturers to improve their devices and customers to choose the best device for their respective needs. The test domains represent requirements in routine pathology as well as education and research. Nevertheless, vendors as well as pathologists requested additional domains, e.g., to determine capabilities of workflow integration, for the 3 rd ISC in 2013.

Design and Implementation of a Digital Pathology Network for Air Force Medical Service Pathology Practice

Nicholas Lancia 1 , Jonhan Ho 2 , Leslie Anthony 3 , Orly Aridor 3 , David W. Glinski 3 , Chris Saylor 3 , Ricky Bond 3 , Joseph P. Pelletier 4 , Dale M. Selby 4 , Emily Green 5 , Kyle Rickard 6 , Anil V. Parwani 7

1
Department of Pathology, 81 st Medical Group Hospital, Keesler Air Force Base, Biloxi, MS, 2 Department of Dermatopathology, University of Pittsburgh Medical Center, Pittsburgh, PA, 3 Office of Sponsored Programs and Research Support, University of Pittsburgh Medical Center, Pittsburgh, PA, 4 Department of Pathology, Wilford Hall Ambulatory Surgical Center, Lackland Air Force Base, San Antonio, TX, 5 Department of Pathology, David Grant Medical Center, Travis Air Force Base, Fairfield, CA, 6 Department of Pathology, Wright-Patterson Medical Center, Wright-Patterson Air Force Base, Fairborn, OH, 7 Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA.

E-mail: Nicholas.lancia.2@us.af.mil

Content

Due to recent advances in digital pathology (DP) and the resulting anticipated changes in pathology practice, the Air Force Medical Service (AFMS) is exploring ways to introduce DP into its pathology labs.

Technology

A model DP network composed of commercially available whole slide imaging (WSI) scanners and their supporting software will be designed. The network should allow sharing of digital images between several geographically remote pathology labs via the AFMS IT network.

Design

To support the design of a small scale model DP network, the unique needs and requirements of AFMS pathologists and pathology system were identified using the contextual inquiry method. Findings were utilized to determine work practices and clinical applications that will most benefit from DP implementation. Commercially available and serviced WSI scanners in the US and supporting software were reviewed. The selected and preferred WSI scanner/system had to support the relevant clinical application/workflows and be compatible with AFMS IT-related infrastructure.

Results

Contextual inquiry revealed that DP will be highly beneficial for AFMS mainly due to the large number and global distribution of its patient population and medical facilities and its unique pathologist/histotechnologist staffing pipeline. Clinical applications and workflows that would most readily benefit from DP implementation were consultations, quality assurance, and global workload distribution. WSI scanner evaluation criteria included a variety of objective technical features. Subjective criteria such as image quality of identical preselected glass slides scanned by each scanner/system and usability of each system were compared. The selected preferred DP system was purchased and installed at four regional AFMS pathology centers. Training processes to support scanning and digital image review were designed, as well as surveys to track pathologists' adoption and use of DP. Following a successful connection of the WSI systems to the AFMS IT network, implementation of a clinical application such as a virtual dermatopathology consultation program within AFMS will be established.

Conclusion

Introduction of new technology such as DP into a large healthcare organization such as AFMS requires careful design and implementation.

Computer-Assisted Analysis of Cytologists' Exploration of Whole Slide Images

Liron Pantanowitz 1,2 , Walid E. Khalbuss 1 , Eugene Tseytlin 2 , Sara E. Monaco 1 , Chengquan Zhao 1 , Jackie Cuda 1 , Anil V. Parwani 1 , Claudia Mello-Thoms 2

1
Department of Pathology, University of Pittsburgh Medical Center, 2 Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, PA.

E-mail: pantanowitzl@upmc.edu

Content

Current clinical practice in cytology involves a two-step process where slides are initially screened by a cytotechnologist and then sent for final review and diagnosis by a cytopathologist. With pathology's migration toward a digital environment, it is important to determine whether these screening and diagnostic tasks can be performed as well on digital slides as they are with glass slides.

Technology

A custom-designed virtual microscope was used in this study. Cytologists had the ability to zoom in on the slides (up to ×40), pan, report diagnostic criteria, and ultimate diagnosis (as well as differential). All interactions with the interface were time-stamped and recorded.

Design

Three cytopathologists and five cytotechnologists voluntarily participated in this study. They read a case set of 12 cytology digital slides, of which 5 depicted Pap tests (conventional and liquid-based specimens), and 7 depicted non-gynecological samples, including 3 FNAs of lung, 1 FNA of liver, 1 FNA of thyroid, 1 FNA of breast, and 1 urine sample. Diagnoses covered a wide range of conditions, from benign (such as infections) to neoplasms (carcinoma, melanoma). Specialized software was developed to analyze the slide exploration strategy of cytologists. Exploration was classified by magnification range into "low" (×1 < mag <= ×10), "medium" (×10 < mag <= ×20) and "high" (×20 < mag <= ×40).

Results

Cytotechnologists took significantly longer to read the digital slides (average 11 min 13 s) than cytopathologists (4 min 51 s) (z = −11.628, P <0.0001). On average, cytologists sampled only about 66% of the entire slide at low magnification levels, and their exploration became even more "economical" as the magnification level was raised. Greater exploration of the slides at low magnification was statistically significantly correlated with correct diagnoses for cytotechnologists (for correct diagnosis, average coverage was 81%, whereas for incorrect diagnosis it was 65%, z = −2.52, P = 0.0117). A high agreement among cytopathologists in the areas of the slide explored was also observed.

Conclusion

Our study suggests that cytotechnologists may have trouble with place keeping using whole slide images, and as a result, they cover less than ¾ of the slide at low magnifications, and this has significant screening and diagnostic consequences. They also take significantly longer to use this medium, which could potentially have a significant impact on workflow.

Recommendations for Validating Whole Slide Imaging for Diagnostic Purposes in Pathology: College of American Pathologists Pathology and Laboratory Quality Center

John H. Sinard 1 , Walter H. Henricks 2 , Alexis B. Carter 3 , Lydia Contis 4 , Bruce A. Beckwith 5 , Andrew J. Evans 6 , Christopher N. Otis 7 , Lisa A. Fatheree 8 , Avtar Lal 9 , Anil V. Parwani 4 , Liron Pantanowitz 4

1
Department of Pathology, Yale University School of Medicine, New Haven, CT, 2 Pathology and Laboratory Medicine Institute, Cleveland Clinic, Cleveland, OH, 3 Department of Pathology and Laboratory Medicine, Emory University, Atlanta, GA, 4 Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA, 5 North Shore Medical Center Salem Hospital, Salem, MA, 6 Laboratory Medicine Program, University Health Network, Toronto, ON, Canada, 7 Department of Pathology, Baystate Medical Center, Tufts University School of Medicine, Springfield, MA, 8 College of American Pathologists, Northfield, IL, 9 University Hospital, London Health Science Center, London, ON, Canada. E-mail: john.sinard@yale.edu

Content

Whole slide imaging (WSI) is increasingly being used for diagnostic purposes, education, and research. Concern has arisen whether WSI can replace the conventional light microscope as the method by which pathologists view patient samples and render diagnoses (primary and/or consultation). Validation of WSI is important to ensure that diagnoses rendered using digitized slides are as accurate as those obtained using glass slides. There are currently no standardized guidelines regarding validation of WSI for clinical diagnostic use.

Technology

In June 2010, the College of American Pathologists Pathology and Laboratory Quality Center convened a non-vendor panel from North America with expertise in digital pathology. The project scope was defined as: "To recommend validation requirements for WSI used for diagnostic purposes."

Design

A computerized search of published articles related to WSI was conducted. Seven hundred and sixty-seven studies met the search term requirements. Twenty-seven studies underwent detailed data extraction to capture evidence in support of validation recommendations. Assessment for strength of evidence, consisting of level of evidence, quantity, size of the effect, statistical precision, and quality assessment (risk of bias) was performed. The panel convened eight times via conference calls and developed draft recommendations. These were made available for open comment in July 2011 in the form of a survey. Approximately 132 responses for each recommendation plus 531 comments were received over the 4-week comment period. The panel had 10 additional conference calls to review the feedback, grade the evidence, and finalize the draft recommendations.

Results

Evidence-based review suggests that diagnoses made using WSI can be as accurate as using glass slides. Validation of the entire WSI system is important prior to using the system for any clinical purpose. The panel developed 12 final recommendations for the validation of WSI systems.

Conclusions

Validation of WSI is necessary to ensure that pathologists using this technique to view digitized glass slides can consistently make the same clinical interpretation as they would from viewing the glass slides using a traditional bright field microscope. Validation should address both technical and interpretative components, and must be specific for the intended clinical use.

Implementation of an Enterprise-Wide Picture Archiving and Communication System for Henry Ford Health System Department of Pathology and Laboratory Medicine

Robert Stapp, Mehrvash Haghighi, Tina Caruana, Ron Brown, J. Mark Tuthill

Department of Pathology and Laboratory Medicine, Henry Ford Hospital, Detroit, MI. E-mail: rstapp1@hfhs.org

Content

We implemented an enterprise Picture Archiving and Communication System (PACS) throughout the laboratories of Henry Ford Health System (HFHS). PACS was deployed across HFHS-Pathology and Laboratory Medicine (PALM) that includes four hospitals. A database server was deployed in our data center with client software installed on workstations for pathologists, grossing, and autopsies.

Technology

The software (Apollo EPMM® v9.4.3, Falls Church, VA, USA) is a file agnostic image database responsible for managing a variety of image acquisition devices based upon the Microsoft.NET® framework and allows secure storage and distribution of digital files acquired from slides, gross specimens, X-rays, whole-slide imaging, electron microscopy, as well as document

scanners. EPMM is capable of interfacing with information systems through its use of HL7, DICOM, and ADT interfaces. Database images can be accessed though the use of the client software or via a web-portal interface.

Design

Staff were directed to acquire images in real-time during the grossing and sign-out process as well as ad hoc additions to cases. Legacy data was converted from our previous archive, requiring importing demographic information of 290,000 cases into the database. This was accomplished by extracting data from our anatomic pathology laboratory information system (LIS); the achieved files, approximately 512,000, were then matched and imported based upon case accession number. In the current live implementation of PACS, demographics are obtained through an ADT interface with our LIS.

Results

Prior to the PACS, all files were hosted on a shared server based on a traditional file-folder hierarchy using accession number as the root for cases. This system became increasingly inefficient due to the volume of cases and resultant slow access time, limited controls to associate the correct image data to the correct patient, and lack of user and role-based security. PACS provided improvements in speed, efficiency, and security. Inter hospital consultations are now performed in real-time. PACS is used to acquire all scanned requisitions and gross images through using automated file mover services. Tumor board presentations have become more efficient and accessible.

Conclusion

PACS system has shown growth in both usage and implementation with increases in efficiency, patient safety, security, and accessibility. As additional features are implemented, PACS will prove to be invaluable for pathologists.

Usage of Inexpensive Consumer Electronics in the Three-Dimensional Capture and Visualization of Gross Specimens

Seung Park, Jeffrey Fine

Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA. E-mail: parks3@upmc.edu

Content

The recent explosion in commoditized imaging technologies has allowed the widespread availability of powerful three-dimensional (3D) imaging and scanning technologies. One such device, the Microsoft Kinect, has garnered a large amount of third-party support, currently enabling low-resolution 3D scans of real-world objects in (or near) real time. In this presentation, we investigate possible clinical uses of this technology in the anatomic pathology laboratory.

Technology

A Microsoft Kinect (Microsoft Corporation, Redmond, Washington, USA); a portable PC (HP ENVY 14; Hewlett-Packard Corporation, Palo Alto, California, USA); software RGBDemo (Manctl Labs) and MeshLab (Instituto di Scienza e Tecnologie dell'Informazione, Pisa, Italy).

Design

In conjunction with a rotating platform and printed scan markers, RGBDemo and the Microsoft Kinect were used to capture point cloud and color data from two gross specimens: a mastectomy for breast cancer and a hysterectomy for fibroids. These data were vertex-cleaned and mesh-reconstituted using MeshLab, and made viewable using the same [Figure 1].



Results

3D models were generated for each specimen, each requiring one slow manual rotation (less than 1 min). Each model had 360° visibility, and could be rotated and viewed in 3D space. Post-processing enabled one to clean the models of extraneous artifacts (e.g., the "table" surface). Image quality was somewhat limited by resolution.

Conclusions

We captured near-realtime 3D models of gross specimens using cheap hardware and freely available software. Although resolution-limited, these systems are under active development for other applications; and we anticipate rapid improvement in image quality. Future higher quality 3D images may represent a leap over conventional photography for documenting gross specimens, teaching and for pathologist telepresence. Surgeons may also benefit, particularly those already using 3D surgical systems in the operating room. Finally, physicians may be better able to communicate with patients using these models. The possibilities are endless.


   Electronic Poster Session Top


Session-1

Thursday, October 11, 2012 (2:40 - 3:30 pm)


Presented in the Regency Ballroom B,C,D

Evaluation of Voice Recognition/Report Template Software in Increasing Efficiency in Hematopathology Report Turnaround Time

Samuel Barasch, Erik Ranheim

Departments of Pathology and Laboratory Medicine, University of Wisconsin, Madison, WI. E-mail: sbarasch@uwhealth.org

Content

Evaluation of the implementation of voiceover software from Voicebrook Company in regards to improve faculty, fellow, and resident efficiency.

The evaluation includes collected data on report turnaround time as well as quantitative and qualitative evaluations.

Technology

Voiceover (Voicebrook Corporation, Lake Success, NY, USA) version 4.4.3. Dragon Medical (Nuance Communication Corporation, Burlington, MA, USA) version 10.10.

Design

Bone marrow aspirate specimen evaluation turnaround is evaluated in March 2012 (1 month prior to installation) and May 2012 (1 month following installation). A survey is submitted to residents, attendings, and the hemepath who use the software.

Quantitative questions asked on the survey were as follows:

Please rate from 1 (strongly disagree) to 5 (strongly agree) the following statements:

  • I use the training/correction features of the software to improve voice recognition
  • Templates are helpful
  • I can get help when I have a question
  • Benefits of using the software outweigh the difficulties in learning and using the software.


Qualitative questions asked on the survey were as follows:

  • How has the software positively or negatively affected the workflow of reporting results?
  • How has the software positively or negatively affected your efficiency?


Results

Mean report turnaround time improved from 2.5 days to 2.0 days (P value, 0.001; 95% CI difference in average TAT is 0.17 days to 0.70 days). Distribution of report turnaround time is also skewed to shorter times following software adaptation (histogram for March TAT [Figure 1] and histogram for May TAT [Figure 2]).

Average scores for survey questions are rated from 1 (strongly disagree) to 5 (strongly agree), which are as follows:

  • I use the training/correction features of the software to improve voice recognition-3.7
  • Templates are helpful-5
  • I can get help when I have a question-2.6
  • Benefits of using the software outweigh the difficulties in learning and using the software-3.3.


Qualitatively, most users commented that report turnaround time has improved. Users also noted that the software and templates streamlined generation of a report on a straightforward case. Most users complained about having to type in or dictate CBC data within the marrow aspirate report.





Conclusion

Improvement in reporting turnaround time was evident soon after software implementation. Because of unanimous consent regarding the usefulness of templates, it is important that a pathologist with appropriate academic and organizational authority be involved with template creation prior to a "go live" on this software. A "super-user" should be designated at the organizational level to coordinate usage and orient new users to templates. Despite an effective vendor training program, correcting errors in voice recognition has not yet been strongly employed by users; though some made qualitative comments about using this in the future. Comments about increased data entry on the part of the pathologist/resident/fellow and concern for concomitant reporting errors are likely to be alleviated through increased integration with the clinical laboratory information system (LIS).

Modeling Morphology in Adeno and Squamous Cell Carcinomas of the Lung

Hyun Ju Choi 1 , David A. Gutman 1,2 , Madhusmita Behera 3 , Lee A. D. Cooper 2 , Gabriel Sica 3 , Taofeek K. Owonikoko 3 , Suresh S. Ramalingam 3 , Fadlo R. Khuri 3 , Joel H. Saltz 1,2

1
Department of Biomedical Informatics, 2 Center for Comprehensive Informatics, 3 Department of Hematology and Medical Oncology, Emory University, Atlanta, GA. E-mail: hyunju.choi@emory.edu

Content

Lung cancer is the leading cause of cancer mortality in the USA. We have developed a computational pipeline to characterize cellular morphology in whole-slide images of lung tumor sections. In studies of glioblastoma, we have observed that morphological patterns are capable of predicting outcome and alterations in molecular cancer pathways (Cooper, JAMIA 2012). To demonstrate the descriptive power of the lung pipeline, we performed preliminary experiments to discriminate between squamous cell carcinomas and adenocarcinomas.

Technology

To segment nuclei, color deconvolution was first applied to separate hematoxylin and eosin signals. Morphological reconstruction was then applied to the hematoxylin signal for denoising. Nuclei were identified in the reconstruction result by Otsu thresholding, and clumped nuclei were separated using a watershed algorithm. Following segmentation, 54 features describing morphometry and texture were calculated for each nucleus. After stepwise feature selection, a support-vector-machine with Gaussian kernel embedding was investigated to distinguish histological subtypes.

Design

Our study included 30 samples each of adenocarcinoma and squamous cell carcinoma of the lung. Neoplastic regions were manually marked by a pathologist. Nuclei in these regions were segmented and the mean nuclear appearance was calculated for each slide. After feature selection, classifiers were trained and evaluated using leave-one-out cross-validation.

Results

The overall classifier accuracy was 92.6%, 87.0% for adenocarcinoma and 98.2% for squamous cell carcinoma.

Conclusion

Computational morphology is a promising area of research that has the potential to improve patient care. Our goal is to develop image-based approaches to recognize finer sub-classifications of tumors that will aid personalization of therapies, and to enable the discovery of morphological hallmarks of specific, predictive genomic alterations. Our preliminary results achieved high degree of accuracy in distinguishing adenocarcinoma and squamous cell carcinomas, suggesting that the extracted features express biologically meaningful information. Future studies will focus on investigating the linkage between tumor cell morphology, patient outcome, and genomics in larger data sets.

Impact of Internet Connectivity on International Telepathology Consultation Service

Ioan Cucoranu 1 , Park Seung 1 , Anil V. Parwani 1 , William Cable 2 , Jeff McHugh 2 , Liron Pantanowitz 1

1
Department of Pathology, 2 Information Services Division, University of Pittsburgh Medical Center, Pittsburgh, PA. E-mail: cucoranuic@upmc.edu

Content

Telepathology is an efficient way of distributing professional expertise over a broad geographic area. The use of telepathology tools has consequently allowed pathology consultation networks to be exploited to provide access to pathology subspecialty expertise in a cost-effective manner. Our institution developed a digital pathology consultation portal to receive digitized cases from local and international clients. The success of this telepathology service depends upon several factors, including the bandwidth of the telecommunication link between clients and consultants. Our goal was to evaluate the impact that Internet connectivity has on our teleconsultation practice.

Technology

A custom developed web-based telepathology portal was used to remotely view digitized slides scanned with a Hamamatsu NanoZoomer 2.0-HT scanner (Bridgewater, New Jersey, USA) and stored on a server in Guangzhou, China. Speedtest.net was used for bandwidth evaluation.

Design

Bandwidth tests for connections between four different workstation configurations (PC versus Mac, Ethernet versus WiFi, 10 Mbps versus 100 Mbps connectivity) at our institution, and various international (China and England), national, and regional servers were performed. We also evaluated whole slide image loading times and turnaround time for digital teleconsultation cases signed-out during a 10-month period.



Results

During this time period, 59 consultation cases were signed-out by subspecialty pathologists at UPMC. Average turnaround time was 3.71 days with 36 cases (61%) being signed-out within 72 h. User experience was satisfactory. Average image loading time was 7 s for digitized slides hosted on servers in China versus 2 s for slides stored on our local imaging servers. Bandwidth tests revealed significantly decreased download and upload speeds for connections to servers in China compared to those located locally [Figure 1]. Various computer configurations did not appear to alter image loading times or quality.

Conclusion

This study shows that it is feasible to provide secondary subspecialty pathology teleconsultation for clients in China in a timely manner for various user workstation configurations using the Internet. Although image loading times for digital files stored in China were slightly longer, and network connection speeds to these remote servers in China was lower compared to that of local servers, this did not negatively impact our sign-out process.

QR Codes (2D Barcodes) in Anatomical Pathology Reports: Proof of Concept

Ioan Cucoranu 1 , Tony Piccoli 2 , Ralph Anderson 2 , Samuel A. Yousem 1 , Liron Pantanowitz 1 , Anil V. Parwani 1 , Douglas J. Hartman 1

1
Department of Pathology, 1 Information Services Division, University of Pittsburgh Medical Center, Pittsburgh, PA. E-mail: cucoranuic@upmc.edu

Content

Quick response (QR) codes are barcodes with two-dimensional patterns that yield many possible combinations. Although traditional barcodes can store 20 digits, QR codes can store up to 7089 digits or 4296 alphanumeric characters. Scanning QR codes with mobile devices such as smartphones with built-in digital cameras facilitates quick reference to online content and even small automation tasks such as dialing phone numbers or sending text messages. In medicine, the use of QR codes to date has limitations. For example, clinicians are using QR codes to help patients manage their conditions, keep their appointments, or for quick access to a patients' medical records. However, to the best of our knowledge, QR codes have not been used in pathology. The aim of this study was to demonstrate that QR codes could be used successfully in surgical pathology reports to direct patients and/or their healthcare providers to appropriate supportive resources.

Technology

Cerner CoPath v.3.2 with PICSPlus (Cerner Corporation, Kansas City, Missouri, USA) and online QR code generator: http://qrcode.kaywa.com (Kaywa, San Francisco, CA, USA).

Design

QR codes, storing uniform resource locators for websites with useful information for patients were generated as PNG files and introduced into Copath PICSPlus as JPEG image files. This allows for the appropriate QR code images to be embedded into the synoptic report field of the pathology report as JPG 2000 files (~1.50 kilobytes). For this proof of concept, we utilized uniform resource locators for our institutions cancer center organ-specific cancer webpages.

Results

It is technically possible to embed QR codes into anatomical pathology reports designed to direct readers of these reports to the appropriate cancer center at our institution [Figure 1].

Conclusion

Embedding QR codes into surgical pathology reports is technically feasible and a novel means of providing patients and their providers with a value-added surgical pathology report with direction to clinical care specialists. QR codes can store more information than traditional barcodes and can be easily scanned at high speed in any direction or orientation with mobile devices. Hard linking with QR codes in paper reports could be replaced by hyperlinks in electronic reports. Although this low-cost technology is easy to implement and use, ongoing maintenance of a QR code library is necessary to ensure that the referred content remains current. The added value of incorporating QR codes into pathology reports is not only for assisting with continuity of care, and contributing to the meaningful use of electronic medical record but also related to the fact that they are useful tools for marketing and patient education.



A Pattern Recognition System for Identifying Cancer Cells in Serous Effusions

David Friedrich 1 , Chen Jin 2 , Yu Zhang, Dipl.-Inf. 1 , Chen Demin 2 , Li Yuan 2 , Leonid Berynskyy 3 , Stefan Biesterfeld 3 , Dietrich Meyer-Ebrecht 1 , Alfred Boecking 3

1
Institute of Imaging and Computer Vision, RWTH, Aachen University, Aachen, Nordrhein Westfalen, Germany, 2 Motic China Group Co. Ltd., Torch Hi-Tech Industrial Development Zone, Xiamen, Peoples Republic of China, 3 Department of Pathology, Division Cytopathology, H.-H.-University Düsseldorf, Germany. E-mail: David.Friedrich@LfB.RWTH-Aachen.de

Content

About 45% of body cavity effusions are caused by metastasizing cancers. For diagnostic assessment, effusion fluid is acquired by aspiration and investigated cytologically. One such cytological technique is DNA Image Cytometry (DNA-ICM), where the diagnosis is based on the DNA distribution measured on morphologically suspicious nuclei. Its diagnostic performance is superior to the conventional interpretation of the cell morphology (Sensitivity 76% vs. 56%, specificity 100% vs. 97.5%). However, more than 40 min are required for manual DNA-measurement and a skilled personnel is often missing. Therefore, the use of DNA-ICM has been limited to a few specialized institutions.

Technology

In order to make this technique available to a wider range of patients, a concept for performing DNA-ICM on effusions more efficiently is presented. A pattern recognition system has been built up, using features describing nuclear morphology, a Random Forest classifier, and a gold standard database (48.777 Feulgen-stained nuclei from seven effusions). In combination with a motorized microscope and an autofocus system (Motic BA610), this system is used to automatically collect and pre-classify nuclei into six classes (artifacts, suspicious cells, granulocytes, lymphocytes, macrophages, and mesothelial cells). Subsequently, the pathologist verifies the classification in the diagnostically relevant classes and DNA ranges.

Design

The pattern recognition system was evaluated and optimized using the leaving-one-out strategy. Subsequently, the system was applied on a test set from 10 patients with the workflow described above and compared to the corresponding manual DNA-ICM results.

Results

The pattern recognition system achieved an overall correct classification rate of 87%. More importantly, 92% of all morphologically suspicious cells were correctly identified. The achieved classification rate plus the final verification of the diagnostically relevant objects were sufficient to confirm the previous diagnosis in all cases from the test set. Interaction time of the pathological expert was reduced from 40 to 5 min per case.

Conclusions

The presented system successfully identified morphologically suspicious cells among thousands of other nuclei and artifacts. Applied in a workflow which requires little interaction by experts, the DNA-ICM diagnoses from the reference measurements could be confirmed in all cases, whereas the productivity was increased by a factor of 8.

The Construction of an Open Source Whole Slide Imaging Library for Graduate Pathology Education

Christopher A. Garcia, Mohamed E. Salama

Department of Pathology, University of Utah, Salt Lake City, UT. E-mail: chris.garcia@utah.edu

Content

Virtual microscopy and online Whole slide image (WSI) libraries are becoming common and have been shown to be a useful educational resource. There are many options and resources currently available for constructing a WSI library. Due to the amount of available resources, it is important to recognize and prioritize what factors are most important to the users. This will help to dictate what technologies and designs will be employed to achieve the desired end product. The goal in constructing our site was to build a site that would maximize accessibility, editing of content, collaboration, ease of use, and flexibility of organization.

Technology

The University of Utah Digital Pathology Project is a web application that is a modified version of New York University School of Medicine's Virtual Microscope (NYU VM) application (cloud.med.nyu.edu/virtual microscope). The viewer is built on the Google Maps Javascript Application Programming Interface (API, Google Inc., Mountain View California, USA). This required a program that would convert commercially available WSI files to JPEG and a compiler to present the files in Google Maps format (written in Python, NYU VM). The site is built on the Django web framework (www.djangoproject.com), which is an open-source project written with Python (www.python.org) programming language. HTML 5, HTML, CSS, and JavaScript (jQuery) were used for client side scripting.

Design

The core application development was completed in the available release of the NYU VM (Beta 5). Modifications to the WSI installer script, metadata framework, and graphical presentation were made in order to customize the application to our workflow and our educational needs. Some functionalities (testing and modes of navigation) were taken out of the current build until it could meet expectations. The user interface was modified to be more navigable by tablets, to be more attractive, and to meet university identity standards. WSI scanning was modified to remove all identifiable information. Standard operating procedures were put in place for scanning, selection of material, uploading files, and editing of files. Instructional videos were created for the use and modification of the site.

Results

The homepage displays a brief introduction and allows the user to immediately view a random case or to search by study set, organ system, or any word in the slide metadata. Each WSI represents a case that can be modified with the addition of images (gross photos, any.jpg, or.png) or metadata (which includes descriptive markers, case history, diagnosis, diagnostic description, and more). The organization of the site is dynamic, with no coding needed to assign a WSI to a different or additional study set. Additionally, no coding is necessary to add or edit metadata for a slide. The WSI set is growing, but currently contains 300 slides. The site can be accessed at imageserver.path.utah.edu. The site fully functions on any browser that supports HTML 5 and is optimized for tablet use. Anyone can view the slides, but you must log in and have appropriate permissions to edit or add cases. The application source code will be made available under the open-source MIT License.

Conclusion

This online WSI library application was built with open-source software and was modified to meet the goals and requirements of our department. It is possible that departments with similar or less information technology resources can adopt and appropriate the same software to meet their needs.

Introduction of a Novel Multivariate Delta Check Approach to Identify Specimen I Inaccuracies

Thomas Kampfrath, James J. Miller

Department of Pathology, University of Louisville, Louisville, KY. E-mail: t0kamp01@louisville.edu

Content

Inaccuracies in specimens of patients may lead to misdiagnosis and inappropriate therapy. The primary purpose of a delta check is to detect misidentified specimens. The traditional delta check procedure compares the change in concentration of an analyte with a delta check limit (DCL) for that analyte. A change greater than the DCL sets a delta check flag for that analyte. This procedure is typically applied to several analytes on a panel. One or more flagged analytes on the panel suggest the possibility that one of the two specimens was misidentified. This procedure is called Univariate (Multianalyte) Delta Check (UDC) and generates many false positive flags. Any improvement would be desirable. Here, we hypothesized that a Multivariate Delta Check (MDC) procedure would produce fewer false-positive flags than the traditional method.

Results

A panel of results on a sample can be graphed as a point in space with one axis for each analyte. A subsequent panel of results can be graphed as a second point in space. If eight analytes are included, there would be eight axes in eight-space indicated as MDC8; with five analytes MDC5. The Multivariate Delta (MD), the distance between these two points, is given by:

MD = SQRT (((analyte1 2 -analyte1 1 ) 2 /(SD RR 1 )) + ((analyte2 2 -analyte2 1 ) 2 /(SD RR 2 )))

The subscripts 1 and 2 refer to the results on the first and second samples, respectively, and the divisor refers to the standard deviation of the reference range.

We calculated the MD for 1035 pairs of patient results from our laboratory on a Basic Metabolic Panel (BMP; Na + , K + , Cl - , CO 2 , glucose, BUN, creatinine, Ca 2+ ) and for 1034 pairs of intentionally misidentified BMP results. Then, we calculated the sensitivity and specificity at various MD values and constructed Receiver-Operating Characteristic (ROC) curves. The UDC5 is currently used by our hospital laboratory and includes the following five analytes (DCL): BUN (10 mg/dL), Ca 2+ (1.5 mg/dL), creatinine (1.5 mg/dL), K + (1.2 mmol/L), and Na + (8 mmol/L). Here, we compare the sensitivities of the MDC8, MDC7 (minus glucose), and MDC5 (same analytes as current UDC5) at the specificity of the UDC5 (0.862).

The sensitivity and specificity of UDC5 is 0.573 and 0.862, respectively. Areas under the ROC curve for MDC8, MDC7, and MDC5 are 0.823, 0.855, and 0.843, respectively. The sensitivities of UDC5, MDC8, MDC7, and MDC5 at a specificity of 0.862 are: 0.573, 0.545, 0.671, and 0.656, respectively. This novel method can be extended to an unlimited number of analytes. Thus, at the specificity of UDC5, the sensitivity of MDC8 was not as high, but the sensitivities of MDC7 and MDC5 were higher.

Conclusion

The MDC7 (BMP without glucose) procedure had the highest area under the ROC curve and higher sensitivity at the specificity of the current UDC5 procedure. The MDC method is more efficient than the UDC method used in our laboratory, which results in more true-positive flags and fewer false-positive flags.

Role of Transcription Factor Th-POK as CD8 Transcriptional Repressor in CD4+ CD8+ Thymocytes Committed to the CD4 T-Cell Lineage

Rahul G. Matnani

Department of Pathology and Laboratory Medicine, Chandler Medical Center, University of Kentucky, Lexington, KY. E-mail: rgmatn2@uky.edu

Content

Developing double-positive (DP) CD4+CD8+ T cells differentiate in thymus and adopt either the single positive (SP) CD4+ helper T cell or CD8+ cytotoxic T-cell phenotype. Th-POK (T helper inducing POZ/Kruppel-like factor) or also called cKrox is expressed specifically in thymocytes committed to CD4 T-cell lineage. Helper-deficient (HD) mice have a point mutation in Th-POK protein and lack CD4 T cells in periphery. Th-POK expression has been found to be necessary to induce gene expression program for CD4 T-cell lineage and inhibit CD8 gene expression program. Th-POK protein has an amino-terminal BTB/POZ domain and four C2H2-type zinc finger domains. The amino-terminal BTB/POZ domain is known to interact with chromatin remodeling factors to mediate transcriptional repression. Th-POK protein is also known to bind to the promoter of Col1a gene in the skin to mediate transcriptional repression. This suggests that Th-POK is likely to play the role of a transcriptional repressor during the process of commitment of DP thymocyte to CD4 T-cell lineage.

Technology

E8 III enhancer region in the CD8 gene is a DP thymocyte-specific enhancer. Using Vector NTI software, known Th-POK consensus binding sequences (5'-GGGAGGG-3', 5'-GGAGGG-3') were identified in the E8 III genomic sequence. The sequence for E8 III enhancer region is from MGI database ID-88347 for Cd8b gene.

Design

To prove in vitro binding, double-stranded synthesized and 32 P end-labeled oligonucleotides from Cd8 loci, which have the Th-POK binding motif, will be initially utilized as probes for EMSA analysis. To prove in vivo binding, chromatin immunoprecipitation (ChIP) analysis will be performed to confirm binding of Th-POK to the enhancer regions.

Results

Using VECTOR NTI software, nine potential Th-POK binding sites were identified in 4.1 kb EcoRI/BamHI E8 III enhancer region. Three representative sense strand sequence of oligonucleotides are shown below with translational start site (+1) shown in parentheses. The bold sequences represent the Th-POK binding motifs and the underlined sequences represent motif on the complementary strand.

1. 5'TCCTTTCCTCCCTCCCCACTTCTC 3' (+6985 to +7008 b.p.)

2. 5'ATTCACTTCCTCCCTCATGTGTGGAGGGCTT TCTAG 3' (+7844 to +7879 b.p.)

3. 5'TGACTAAAGGGGGGGGGGGTGCATTTG 3' (+8424 to +8450 b.p.)

Conclusion

The experiments proposed in this study will help to determine the role of Th-POK in silencing of CD8 loci in positively selected DP thymocytes committed to CD4 T-cell lineage. This can result in therapies directed to increase CD4 T-cell count in HIV and in other immunocompromised patients.

Analysis of Whole Slide Imaging Scan Failures

Tushar N. Patel 1 , Seung Park 2 , Jon Duboy 2 , Anil V. Parwani 2 , Walid E. Khalbuss 2 , Liron Pantanowitz 2

Departments of Pathology, 1 Rush University, Chicago, IL, 2 University of Pittsburgh Medical Center, Pittsburgh, PA. E-mail: tushar_patel@rush.edu

Content

Digitizing glass slides using whole slide imaging (WSI) scanners involves slide preparation and rapid, automated image acquisition aimed at generating an accurate digital representation of the original glass slide. The aim of this study was to evaluate the frequency and reasons for WSI scan failures at an academic institution.

Technology

Aperio ScanScope XT 120 WSI scanner and HP xw4600 computer workstation (2.66GHz Intel Core 2 Quad Q9400, 4GB RAM).

Design

During a 7-month period, 4262 glass slides were digitized using a WSI scanner for various reasons (archiving, quality assurance, education, and research). Prior to scanning, slides were cleaned and correctness of cover-slipping was ensured. Double-thick previously repaired slides were excluded. All slides that failed scanning during this period were recorded as WSI failures. Slides that failed initial automatic scanning were manually re-examined at high magnification, directed focus selection performed and image re-acquisition attempted.

Results

A total of 19 slides (0.45%) were considered WSI failures. Thirteen (68%) of these failures were non-gynecological cytology specimens, all of which had insufficient tissue (too little or pale) to permit autofocusing by the WSI scanner. Attempts to manually rescan these slides were unsuccessful due to insufficient focus points and image stitch mismatching. Five (26%) glass slides of a neuropathology case from 1960 were too thick to be scanned. One slide (5%) was received irreparably broken and did not scan. Technical issues and user error did not result in any WSI failures.

Conclusions

WSI scanning technology continues to develop. Nonetheless, as these data show, a small percentage of glass slides may fail to scan, despite proper glass slide preparation prior to image acquisition. Insufficient tissue, non-standard or broken glass slides may contribute to WSI failures. Such WSI failures need to be taken into consideration when adopting digital pathology. Further research is needed to identify technological and workflow solutions to minimize WSI scan failures, as well as improve tissue detection and autofocusing capabilities of WSI scanners.

Implementation of an Enterprise-wide Picture Archiving and Communication System for Henry Ford Health System Department of Pathology and Laboratory Medicine

Robert Stapp, Mehrvash Haghighi, Tina Caruana, Ron Brown, J. Mark Tuthill

Department of Pathology, Henry Ford Hospital, Detroit, MI. E-mail: rstapp1@hfhs.org

Content

We implemented an enterprise Picture Archiving and Communication System (PACS) throughout the laboratories of Henry Ford Health System (HFHS). PACS was deployed across HFHS-Department of Pathology and Laboratory Medicine (PALM) that includes four hospitals. A database server was deployed in our data center with client software installed on workstations for pathologists, grossing, and autopsies.

Technology

The software (Apollo EPMM® v9.4.3, Falls Church, VA, USA) is a file agnostic image database responsible for managing a variety of image acquisition devices based upon the Microsoft.NET® framework and allows secure storage and distribution of digital files acquired from slides, gross specimens, X-rays, whole-slide imaging, electron microscopy, as well as document scanners. EPMM is capable of interfacing with information systems through its use of HL7, DICOM, and ADT interfaces. Database images can be accessed through the use of the client software or via a web-portal interface.

Design

Staff were directed to acquire images in real-time during grossing and sign-out process as well as ad hoc additions to cases. Legacy data were converted from our previous archive, requiring importing demographic information of 290,000 cases into the database. This was accomplished by extracting data from our anatomic pathology laboratory information system (LIS); the achieved files, approximately 512,000, were then matched and imported based upon case accession number. In the current live implementation of PACS, demographics are obtained through an ADT interface with our LIS.

Results

Prior to the PACS, all files were hosted on a shared server based on a traditional file-folder hierarchy using accession number as the root for cases. This system became increasingly inefficient due to the volume of cases and resultant slow access time, limited controls to associate the correct image data to the correct patient, and lack of user and role-based security. PACS provided improvements in speed, efficiency, and security. Inter-hospital consultations are now performed in real-time. PACS is used to acquire all scanned requisitions and gross images through automated file mover services. Tumor board presentations have become more efficient and accessible.

Conclusions

PACS system has shown growth in both usage and implementation with increases in efficiency, patient safety, security, and accessibility. As additional features are implemented, PACS will prove to be an invaluable for pathologists.

HPath: A Browser-Based, User-Extensible Report Generator for Clinical Environments

Sean R. Wilkinson, Jonas S. Almeida, James R. Hackney, Benjamin C. Hill

Department of Pathology, University of Alabama at Birmingham, Birmingham, AL. E-mail: wilkinson@uab.edu

Content

This presentation describes HPath, a user-extensible tool for generating text reports from templates and provides an example for creating hematopathology reports, which typically follow a well-structured format suitable for automatic generation. HPath allows domain experts to quickly generate text reports using interactive form elements within a web browser. The generated text can be edited within the tool if necessary and then copied and pasted into the hospital Laboratory Information System (LIS). The data layer is completely decoupled from the presentation layer: a template system enables clinicians to adapt the tool to their own needs by defining and/or loading other templates. Using HPath requires no installation or special software, and extending it requires minimal programming knowledge.

Technology

HPath runs entirely in the "sandbox" environment of an ordinary web browser: the text report generated does not leave the user's machine and the application does not have programmatic access to the machine's file system.

Design

HPath's template specifies an interactive form for composing a formatted report using only a mouse- or touch-based interface and provides instant visual feedback by re-generating and displaying output dynamically. A template is any HTTP-accessible plain-text document whose contents conform to HPath's domain-specific language (DSL), which is based on the human-readable JavaScript Object Notation (JSON) format.

Results

The source code is made publicly available, with version control and documentation, at https://github.com/wilkinson/hpath. As detailed in the documentation, the default hematopathology template, loaded by appending "?https://raw.github.com/wilkinson/hpath/templates/default.js", can be replaced or extended by additional user-defined modules.

Conclusion

HPath can streamline workflows in hospital environments for which the LIS interface is difficult or impossible to modify, and it can generate a wide variety of reports such as the hematopathology example included as the default template. This report generation tool illustrates uses of web technologies for easier application development and maintenance and allows for collaborative configuration by domain experts.

An Intelligent Spreadsheet to Aid in Bone Marrow Engraftment Analysis

Christopher Williams, S. Terence Dunn, Richard A. Allen

Department of Pathology, University of Oklahoma Health Sciences Center, Oklahoma City, OK. E-mail: christopher-williams@ouhsc.edu

Content

Polymerase chain reaction-based short tandem repeat (STR) analysis, performed for bone marrow engraftment evaluation, is a tedious and time-consuming process and it requires relatively complex, manual calculations. Although these calculations are repetitive, choosing appropriate equations for analysis of data is context-dependent, necessitating some element of decision-making. In addition, noise from aberrant amplification of STR loci needs to be taken into account. The intent of this application is to automate repetitive calculations while using the technologist's expertise to ensure accurate results.

Technology

A prototype of the application was initially developed using Lab View (National Instruments; Austin, TX, USA). The prototype was used as a proof-of-concept and a guide for the final architecture. The three main considerations for the final development environment were cost, ease of software distribution onto university supported computers, and usability. Despite certain drawbacks, Microsoft (Redmond, WA, USA) Excel with Visual Basic for Applications best fulfilled these criteria since it is already licensed and installed on every computer at our institution and end-users are familiar with the program.

Design

Plain text data files formatted as comma separated lists (.csv) are imported directly from an Applied Biosystems (Carlsbad, CA, USA) capillary electrophoresis instrument as spreadsheets, one file each for the recipient (pre-transplant), the donor, and post-transplant samples. Visual Basic subroutines scan these spreadsheets and identify the overlapping peaks (amplified products from STR analysis) between the post-transplant and donor/recipient samples. Rather than simply returning a result, the subroutine returns the chosen formula and resulting calculation to facilitate review and manipulation as needed.

Results

Running historical data through the application produced highly concordant results with manual calculations and required minimal manipulation from the reviewer. The time required to run the application and review the results is considerably less than performing manual calculations, saving approximately one hour per test. Our molecular laboratory has performed a comprehensive validation of the application and has recently deployed it as part of the STR analysis of clinical samples.

Conclusion

The main objective of providing a tool to increase the efficiency of bone marrow engraftment STR analysis was successfully achieved. Although not eliminating the need for operator input, it does eliminate time-consuming and error-prone manual calculations and facilitates more efficient review.

Ensuring Appropriate Reporting of Urgent, Significant, and Unexpected Anatomic Pathology Diagnoses to Patients Using Electronic Transmission of HL7 Flags to the EHR

Joseph A. Zeitouni, Philip Chen

Department of Pathology, University of Miami Miller School of Medicine, Miami, FL. E-mail: jzeitouni@med.miami.edu

Content

The reporting of urgent, significant, and unexpected diagnoses in anatomic pathology as a patient safety issue has been the subject of a recent consensus statement by the College of American Pathologists and the Association of Directors of Anatomic and Surgical Pathology. With widespread adoption of EHR in clinical practices, well-designed communications between AP LIS and EHR offer opportunities for pathologists not only to report these significant findings to treating physicians but also to ensure that appropriate patient management is rendered.

Technology

Copath 6.0 (Sunquest Information Systems, Tucson, Arizona, USA), Epic IU 5 (Epic Systems Corporation, Verona, Wisconsin, USA).

Design

At the time of electronic sign-out in the AP LIS, the pathologist will be presented with a pop-up window with a prompt to flag the case as containing an urgent, significant, or unexpected finding. A web link to a reference of current medical executive committee approved criteria for case flagging will be available. If the pathologist flags the case, a data collection window will appear to allow documentation of the discussion with the physician. Additionally, a flag will be released as part of the HL7 result to the EHR, where an "Urgent/Significant diagnosis encounter" will be opened. The encounter in the EHR must be closed by the responsible physician within 48 h by electronically signing an attestation that the patient was told of his/her diagnosis. Aging encounter reports generated in the EHR will be monitored daily and subjected to an escalation policy to ensure that physician manage the patients in a timely fashion.

Results

Pathology Signout


  1. Answers prompt: Is this an urgent/significant/unexpected pathology result? Yes/No
  2. Documents: "Results reported to: ............", OR "Unable to reach provider."


Copath to EPIC HL7 Interface
"Significant" Flag sent across the interface.

EHR

EHR recognizes the flag, and creates a "Urgent/Significant diagnosis encounter.

Patient Care

  • Physician discusses results with patient must document discussion in EHR within 48 hours.
  • The "Urgent/Significant diagnosis encounter" is closed.


Conclusions

There is a growing recognition that if pathologists are to continue to be regarded as members of the clinical management team, then they must assume responsibilities of ensuring that patients are managed appropriately based on the findings from diagnostic tests or procedures. Pathologists should not be afraid to leverage the EHR to develop systems that improve patient care. By doing so, they extend their influence beyond the four walls of the laboratory, and can re-establish themselves as being physicians at the center of patient care.


   Scientific Session Top


Session-1

Data Models, Bioinformatics and Personalized Medicine

Thursday, October 11, 2012 (7:30 am - 8:55 am)


Regency Ballroom A

ImageJS: Personalized, Participated, Pervasive, and Reproducible Image Bioinformatics in the Web Browser

Jonas S. Almeida, James R. Hackney

Department of Pathology, Division of Informatics, University of Alabama at Birmingham, Birmingham, AL. E-mail: jalmeida@uab.edu

The ImageJS tool reported in the July 2012 issue of Pathology Informatics (J Pathology Informatics 3:25) describes a successful attempt to use a Web App ecosystem approach for the dissemination/deployment of Pathology Informatics applications. This approach, validated there with an image analysis application, only became a realistic proposition with the development of the modern web browser combined with the recent emergence of the third generation of Web technologies (Web 3.0). The key advantages of this approach derive from the browser's reliance on the code migrating to the machine where the access to the data already exists. This decreases the exposure of sensitive data, which no longer leave the protected clinical informatics environment. It also facilitates having multiple partners, for example, computational statisticians, to develop individual modules which the user can then compose as needed for specific applications.

The basic module of ImageJS, which can be obtained by directing the browser to imagejs.org, includes the module orchestration methods that will respond to the URL of additional modules. This illustration can also be visualized in the two webcasts listed in that web page where multiple combinations of modules, for example to calculate cellular proliferation in KI67 labeled images, are assembled into a single link. As a result, a pathologist would only have to direct the browser to a tailored link to assemble an advanced image analysis application configured for a specific problem: with no download or installation. This gives substance to the argument that Web App ecosystems, such as ImageJS, have significant advantages over more conventional approaches, such as NIH's ImageJ, as a vehicle to deploy Pathology Informatics applications.

Content

This presentation describes and expands on the ImageJS tool reported in the July 2012 issue of Pathology Informatics (J Pathology Informatics 3:25). Modern web browsers now include code interpreters and support code distribution architectures that are advantageous as a computational environment to deliver informatics applications. In a nutshell, this presentation explores the use of web Apps to deliver image analysis directly (no downloads or installations) to where it is needed. We will argue, with examples such as determining proliferation from KI67 labeling images, that this approach has important advantages over conventional systems such as NIH's ImageJ.

Technology

Only web technologies that are natively supported by the browser are used (HTML5 and JavaScript), combined with an open architecture that invites distributed application development.

Design

A core module is delivered with the imagejs.org URL with analytical and user interface components being then loaded with their own script tags. The effect is that a particular image analysis solution is automatically configured from a string of URL's, blurring the distinction between a "personalized" application and an analytical workflow.

Results

A configurable Web App ecosystem for image analysis, inviting module development by statisticians and human-machine interface researchers alike, is made available with open source and in the public domain. This result is detailed in a July 2012 Pathology Informatics report, and is also illustrated by a number of YouTube webcasts, including the two listed in http://imagejs.org. This ability of non-disruptively delivering advanced computational statistics applications to the point of care also caught the attention of the media which described it as the "angry birds" approach.

Conclusion

The emergence of the modern web browser coupled with the more open architectures of the third generation of Web technologies creates novel opportunities for pathology informatics. The ImageJS application to image analysis suggests that ecosystems of Web Apps have fundamental advantages as informatics solutions as concerns deployability, configurability, reproducibility, and protection of sensitive personal health information (PHI).

ClusterFASTQ: A Method for the Identification of Translocations in Clinical Next Generation Sequencing Data

Eric J. Duncavage 1 , Haley Abel 2

1
Department of Anatomic and Molecular Pathology, 2 Division of Statistical Genetics, Washington University College of Medicine, St. Louis, MO. E-mail: eduncavage@path.wustl.edu

Content

The identification of gene translocations has both diagnostic and prognostic significance in molecular oncology. Current clinical methods for translocation detection generally rely on low-resolution techniques such as fluorescence in-situ hybridization (FISH), however, recent work has demonstrated that translocations can be identified by targeted next generation sequencing (NGS). Existing methods for translocation detection in NGS data involve the identification of discordant paired-end sequencing reads, however, these methods are subject to a high false-positive rate that makes them unsuitable for clinical molecular diagnostics. ClusterFASTQ improves upon existing methods and offers a greatly reduced false-positive rate while maintaining a high sensitivity.

Technology

ClusterFASTQ is a command line utility implemented in Java. It accepts aligned sequence (BAM) files as input and outputs the chromosomal positions of both translocation partners as well as a contig spanning the breakpoint sequence.

Design

Many existing methods for the detection of translocations in NGS data rely on the identification of discordant paired-end reads where paired reads map to different chromosomal regions. Although such methods are sensitive and computationally efficient, they produce a high number of false-positive results due to repetitive elements in the human genome. Other methods reduce the false-positive rate by looking for chimeric single-end reads, but the performance can be heavily dependent on the choice of aligner, and the methods generally do not produce large breakpoint-spanning contigs. ClusterFASTQ improves upon these methods by first identifying clusters of discordant paired reads, then validating the clusters by re-mapping the unmapped or incompletely mapped partners of nearby reads. Finally, it assembles all reads mapping to the vicinity of the cluster into a contig, typically 300-3000 bp long.

Results

DNA sequences from eight formalin-fixed cancers with known ALK or MLL translocations were analyzed by both Breakdancer and ClusterFASTQ. Both tools correctly identified the translocations, however, Breakdancer produced an average of ~1200 putative translocations/case whereas ClusterFASTQ identified only one/case.

Conclusion

ClusterFASTQ allows for the rapid detection of translocations from targeted NGS data. Using a secondary single-end read re-alignment step, ClusterFASTQ greatly reduces the number of false-positive translocation results obtained, thereby permitting NGS-based translocation detection in the clinical laboratory.

Patients Accessing Laboratory Results via Patient Portal: What are the Risks?

Maggie L. Hopkins 1 , Alexis B. Carter 2

1
Department of Pathology, University of Utah School of Medicine, Salt Lake City, UT, 2 Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA. E-mail: maggielou.hopkins3@gmail.com

Content

Patient portals (PTPRs) are web portals typically offered by a healthcare entity to allow patients to view their entity-related medical information. The use of PTPRs is increasing, in part because of programs established by the US government including meaningful use (MU) of electronic health records. Several objectives of MU Stage 1 and recently finalized MU Stage 2 require that a percent of patients have access to a PTPR that includes lab results. However, controversy exists on whether direct release of medical information exposes patients to risks of misinterpretation or other harm. To answer this question, the authors undertook a review of the existing medical literature.

Technology

PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) and SurveyMonkey (http://www.surveymonkey.com) were used to review the literature and collect data, respectively.

Design

Recent medical literature on PTPRs was retrieved from PubMed by searching for terms in quotes: "patient portal" and "patient portals." A total of 98 articles were retrieved and filtered down to 34, based on the whether the title was relevant to the topic. Subsequently, two independent reviewers reviewed each of the 34 articles. Articles were excluded if their focus was on manual entry of data by the patient (e.g., personal health record), electronic visits or if the full text could not be acquired. After exclusion, 10 articles remained which were categorized according to their focus.

Results

The papers were reviewed for whether or not they covered the following items, and the percent which did address was recorded: Lab results available: 50% (5), ability to access portal: 70% (7), described patient demographics: 50% (5), reported access disparities: 40% (4), described usability: 60% (6), access to sensitive results: 20% (2), compliance with laws or standards: 10% (1), specific vendor platforms: 10% (1) [Table 1].

Conclusions

Peer-reviewed literature indicates that patients are most likely to use PTPLs to view laboratory results. Despite this, peer-reviewed articles examining how patients interpret and use laboratory results are lacking. Limitations of this study include lack of standard terminology and overlap with "personal health record(s)," thereby likely artificially reducing the number of articles for review. Regardless, much more work is needed to determine compliant standards by which laboratory data should be displayed in PTPLs for ease of patient comprehension.



Quantification of Epidermal Growth Factor Receptor Expression in Colorectal Cancer using Image Analysis

Ryan A. J. Hutchinson 1 , Jacqueline A. James 1 , Richard Wilson 1 , Richard A. Adams 2 , B. Jasani 2 , Manuel Salto-Tellez 1 , Peter W. Hamilton 1

1 Centre for Cancer Research and Cell Biology, School of Medicine, Dentistry and Biomedical Science, Queen's University Belfast, Belfast, United Kingdom, 2 Institute of Cancer and Genetics, Section of Oncology and Palliative Medicine, Cardiff University School of Medicine, Cardiff, United Kingdom.

E-mail: rhutchinson09@qub.ac.uk

Content

The COIN trial was a major multinational study examining the comparison of continuous chemotherapy with and without cetuximab to intermittent chemotherapy as a first-line therapy in patients with previously untreated, advanced colorectal cancer. Original findings from the COIN trial suggested that pathology based visual scoring of epidermal growth factor receptor (EGFR) expression using immunohistochemistry (IHC) in advanced colorectal cancer tissue microarrays did not indicate any predictive value for treatment with cetuximab and chemotherapy in first line therapy. The aim of our study was to examine whether automated IHC image analysis has the potential to provide new evidence as to the utility of EGFR IHC as a predictive biomarker.

Technology

The complete set (27 slides) of IHC-stained, advanced colorectal cancer tissue microarray slides was digitized using an Aperio ScanScope CS2 scanner. Representative regions were randomly marked up across 23 tissue microarray cores by the original COIN trial pathologist. A total of 70 regions were annotated which were representative of a range of staining intensities. The digitized, IHC-stained slides were imported into Definiens Tissue Studio for EGFR quantification. This cohort of marked-up tissue microarray cores was used as a training set to develop the algorithm. A region recognition method applied to EGFR IHC images identified histological regions and a novel image analysis solution was created to identify positive membrane staining using subsets of identified regions within the EGFR IHC images.

Results

The defined algorithm was capable of accurately segmenting and measuring positive cell membrane EGFR expression across the range of colorectal cancer samples. This allowed for the successful selection of epithelial tumour cells and exclusion of positivity in stromal/non-tumour regions. A direct comparison between visual IHC scores and computerized image analysis-derived IHC scores showed a strong correlation (r = 0.96, P < 0.0001). Some results suggested that visually marked up cores with a lower histological score were scored higher in some instances using image analysis.

Conclusion

The application of image analysis facilitates the automated quantification of the EGFR IHC expression, providing accurate membrane segmentation and measurement. Although visual and automated IHC evaluation showed strong concordance in this sample set, the clear advantage of image analysis is its reproducibility and consistency across large numbers of samples. Having now established a reliable image analysis method for quantification of EGFR expression in colorectal cancer, future work will examine the role of EGFR image analysis in predicting response to cetuximab in the COIN trial patient cohort.

A Digital Imaging and Communication in Medicine Prototype in XML with Relationships

Robert C. Leif, Stephanie H. Leif

Newport Instruments, San Diego, CA. E-mail: rleif@rleif.com

Content

CytometryML consists of XML schemas that describe the specimen processing, acquisition, and results of cytometric measurements. Presently, these measurements can be performed with a flow cytometer or digital microscope. Because of the considerable overlap between cytometry and digital microscopy, data-types have been incorporated from both the International Society for Advancement of Cytometry (ISAC) and the Digital Imaging and Communication in Medicine (DICOM) standards. The DICOM separation of studies and instances has been included. The architecture of CytometryML now includes the capacity to describe relationships in a manner, which is similar to that of the resource description framework (RDF) and the ZIP file container for both the meta- and binary data and is based on the popular EPUB standard.

Technology

The schemas are written in the XML Schema Definition (XSD1.1) language and validated to demonstrate adherence to XSD1.1. Their content was tested by translating specific XSD elements into XML and filling in the values of the objects contained therein. The use of an element-based implementation of relationships permits bidirectional and multiple relationships between two objects to be expressed.

Design

Modularity of the design was enhanced by basing the schemas on objects. The DICOM hierarchical separation of a metadata-containing series and separate instances, which also include binary data, was maintained. Because of compatibility with draft Supplement 161, CytometryML should be able to read XML data that is stored on the picture archiving system, PACS while minimally loading the server, as well as being compatible with future efforts to write XML data on the PACS.

Results

An XML-based system that incorporates data-types from existing standards and provides enhanced, but simple-to-understand relationships has been created. Preliminary data indicate that these XML data-types can be used with XHTML5, which would permit the creation of a medical informatics system that has access to the full power of the Internet.

Conclusion

CytometryML can be considered as a basis for a collaborative effort between pathologists and cytometrists for development of a continuum of complimentary interoperable standards and a prototype of a future Internet-based version of DICOM, DICOM 4.

Best Block Designation in Surgical Pathology: A Help or a Hindrance to Subsequent Molecular Studies?

Andrew M. Quinn, Frank Kuo

Department of Pathology, Brigham and Women's Hospital, Boston, MA. E-mail: andrew.m.quinn@gmail.com

Content

Selecting blocks with tumor suitable for testing is a time-consuming step in molecular diagnostics workflow. In order to improve turnaround time, our department instituted a quality improvement initiative in 2010 by which surgical pathologists select a "best block" (BB) for molecular testing and record its tumor content during case finalization. The purpose of this study was to compare pre-analytical and analytical processing intervals between molecular cases for which a BB was designated and those for which it was not.

Technology

Two partially automated, real-time tracking systems were employed: an in-house, web-based laboratory information management system for following pre-analytical and analytical molecular case processing time points, and an in-house enhancement to Sunquest PowerPath (Sunquest Information Systems, Tucson, AZ, USA) for BB designation and characterization. Time points were logged via electronic verification within the web-based laboratory information management system. Data analyses were performed via GraphPad Prism (GraphPad Software, Inc., San Diego, CA, USA).

Design

Retrospective review of time points for 2185 consecutive tumor genotyping cases (August, 2011 to July, 2012) at a large academic medical center (Brigham and Women's Hospital, Boston, MA, USA) was performed. Time points include: report review, block/slide adequacy review, pathologist approval for molecular analysis, molecular analysis start and end (DNA extraction and sequencing), and molecular report finalization. Cases were divided into three groups: BB designated and ultimately used (BB used), BB designated but not used (BB not used) and BB not designated (no BB). Non-parametric analysis of variance calculations with post-hoc Dunn's multiple comparison tests were used for comparison with significant P values of less than 0.05.

Results

A summary of the results is presented in [Table 1].

Conclusions
"BB" designation at the time of surgical pathology case finalization results in significant improvement in the overall turnaround time for subsequent molecular analyses. The improvement is best realized in the pre-analytical phase of reviewing a case, but may reflect selection bias. Surprisingly, BB not used cases have significantly shorter pre-analytical intervals than BB used cases, suggesting a confounding variable. The variability of numerous pre-analytical intervals (data not shown) indicates that further refinements and subsequent review of the tracking process is needed for optimization.



Session-2

Workflow

Applied Informatics and Clinical Workflow

Thursday, October 11, 2012 (7:30 am - 8:55 am)


Acapulco Room

How Can Laboratories Help Their Hospital Decrease Readmission Rates for Heart Failure Patients?

Eugenio Zabaleta 1 , John Burgess 1 , Kay Truax 2 , Michael Patterson 3

1 Departments Laboratory, 2 Patient Education, and 3 Administration, MedCentral Health System, Mansfield, OH. E-mail: ezabaleta@medcentral.org

Content

Starting third quarter of 2012 hospitals will experience decreased Medicare reimbursement for all Medicare discharges when their readmission rates are higher than expected according the Hospital Readmissions Reduction Program. In response MedCentral Health system developed heart success, which is a program to assist patients with heart failure (HF) to avoid readmission to the hospital. As well, MedCentral utilizes Information technology (IT) to help identify all HF patients that could benefit with this program.

Technology

To respond to the new health-care challenges, MedCentral has decided to use a Laboratory Decision Support Software called RippleDown® (from Pacific Knowledge Systems) to help identify all HF patients that could benefit with this program. Within this software, rules were developed for the purpose of stratifying patients according to their BNP and Troponin results (Circulation 2003; 108; 833-838). Also, this software will automatically communicate this information to the heart success personnel with minimal obstruction to clinical operation.

Design

Since mid-October 2011, MedCentral Laboratory is able to perform risk stratification on every inpatient who has a BNP test done (BNP is the trigger). Once the risk is calculated, the result will be sent via E-mail with patient demographic information, risk calculation, and other laboratory data to the heart success personnel to prioritize patient management.

Results

The US national heart failure re-admission rate is 24.8% and can be as high as 40% according to CMS (data from 2011). MedCentral hospital rate for people not enrolled in the heart success project is 26.7%, whereas it is 4% for the enrolled patients. This project went live on October 2011 and since then, the hospital average of HF re-admission rate is 20% or below (20.0% in October, 13.3% in November, and 16.1% in December). No data are available for 2012.

Conclusions

Even though most of the patients recruited for heart success are through physician referrals, the automated to Congestive Heart Failure (CHF) patient risk stratification project serves as a safety net for identifying those patients recently diagnosed with CHF or those who are not referred, or for those not treated in our main hospital.

Using Camera Systems to Read Entire Batches of Barcodes and More

Lyman T. Garniss

Partners Healthcare Systems - IT, Massachusetts General Hospital, Boston, MA. E-mail: LGarniss@partners.org

Content

Massachusetts General Hospital (MGH) Pathology Service is now using 2-D barcodes on all cassettes and slides. We are also tracking and routing these materials using Sunquest CoPath SMART. In some areas, it is desirable for patient safety and for workflow efficiency to scan entire batches of cassettes and/or slides instead of scanning individual assets.

Technology

The technology to accomplish this project included A Cognex industrial camera system with software that enables the identification and reading of 2-D barcodes. The Sunquest CoPath SMART system, unique asset identifiers for specimen cassettes and slides encoded on 2-D barcodes; Boston Workstation scripting tools, and an enclosure created by Machine Vision Consulting.

Design

A rack of barcodes cassettes or a folder of slides is loaded into the camera enclosure. The rack or folder activates a switch in the enclosure, the system is activated turning on the lights, and the picture is taken. The picture is transferred to a computer, analysis is performed on the picture and the barcodes and a list of asset numbers is produced and sent to a file.

Results

The file is then read by a Boston WorkStation script and then checked against the Sunquest CoPath database to ensure all specimens are together (none are missing) and the script also "checks" the assets out of the area in the tracking system. The system can complete the entire process of reading, checking, and updating the CoPath database on 78 cassettes in 2 min. Doing this manually takes a human over 10 min and does not allow for ensuring that all cassettes that should be together are together.

Conclusion

Batching reading of pathology assets (cassettes and slides) using an industrial camera system increases patient safety and allows for faster throughput of specimen processing when using tracking and routing software.

Clinical Genomicist Workstation

Rakesh Nagarajan 1 , Mukesh K. Sharma 1 , Joshua Phillips 2 , Saurabh Agarwal 2 , Wesley S. Wiggins 2 , Savita Shrivastava 1 , Sunita B. Koul 1 , Madhurima Bhattacharjee 1 , Caerie D. Houchins 1 , Raghavendra R. Kalakota 1 , Bijoy George 1 , Rekha R. Meyer 1 , David H. Spencer 1 , Christina M. Lockwood 1 , TuDung T. Nguyen 1 , Eric J. Duncavage 1 , Hussam Al-Kateb 1 , Catherine E. Cottrell 1 , Suhasini Godala 2 , Ravi T. Lokineni 2 , Sameer M. Sawant 2 , Vasudev Chatti 2 , Suresh Surampudi 2 , Raja Rao Sunkishala 2 , Ramakant Darbha 2 , Sharath Macharla 2 , Jeffrey D. Milbrandt 3 , Herbert W. Virgin 1 , Robi D. Mitra 3 , Richard D. Head 3 , Shashikant Kulkarni 1 , Andrew Bredemeyer 1 , John D. Pfeifer 1 , Karen Seibert 1

1
Department of Pathology and Immunology, Washington University in St. Louis, St. Louis, MO, 2 SemanticBits LLC, Herndon, VA, 3 Department of Genetics, Washington University in St. Louis, St. Louis, MO. E-mail: rakesh@wustl.edu

Content

The use of NextGen Sequencing clinically necessitates the need for informatics tools that support the complete workflow from sample accessioning to data analysis and reporting. To address this need, we have developed Clinical Genomicist Workstation (CGW).

Technology

Apache 2, Tomcat 6, Grails 2.0, Spring 3.1, Drools 5.2, YUI 2, jQuery 1.7, HTML5, Groovy 1.8, FreeMarker 2.3, Java 6, Oracle Database 11g, Red Hat Enterprise MRG 7.6.3, Integrated Genomics Viewer 2.1.

Design

CGW is a secure, n-tiered application. A web browser submits requests to application servers that persist data in a relational database. CGW utilizes Grail and Groovy, permitting advanced, automated, and customized access to the application's features. RedHat MRG is used to launch and track complex analysis pipelines.

Results

CGW is used by Washington University Genomic and Pathology Services for clinical genomic testing of many cancers. Orders are accessioned in CoPath and subsequently, the CGW. DNA from specimens is subjected to NextGen sequencing, which is tracked in the CGW. Informatics pipelines for variant detection are executed from the CGW once the sequencing run completes. Results are loaded into the CGW database following analysis completion. Called variants are viewed in context of data from other variant databases such as dbSNP, COSMIC, HGMD, and ClinVar, and previously curated interpretations for variants are automatically inserted into a draft clinical report. The draft report is reviewed and finalized by the clinical genomicist and messaged to the medical record for persistence. CGW has been used to sign out over 100 cases since November 2011, and more than 2000 cases are expected to be signed off by June 2013. There are 22 ordering oncologists and 7 clinical genomicists that use the CGW. Future plans include increasing the number of comprehensive cancer panel genes from 27 to 50 and introducing test panels for drug metabolism and transport, renal, and cardiac diseases. Finally, CGW is in the process of being adopted at several other clinical genomic laboratories nationally and globally.

Conclusion

CGW is a "soup-to-nuts" software solution for managing clinical genomic testing through order intake, sequencing, analysis, case review, interpretation, and reporting.

Development of Automated Testing Solution for Anatomic Pathology Laboratory Information System: Background, Technical Design, and Future Vision

J. Mark Tuthill 1 , Ron Brown 1 , Jennifer Lyle 2 , Chris Gardenhour

1 Division of Pathology Informatics, Department of Pathology and Laboratory Medicine, Henry Ford Hospital, Detroit, MI, 2 Software Testing Solutions, Tucson, AZ. E-mail: mtuthil1@hrhs.org

Content

We have previously reported on use of an automated testing solution for clinical pathology CLIA validation testing. Working with our vendor partners, we developed a novel testing solution to support validation testing of our CoPathPlus Laboratory Information System (LIS), specifically for the surgical pathology cases. Although validation testing is less complex for anatomic pathology laboratory information system (AP-LIS), functional testing is becoming increasingly complicated as AP-LIS increase in sophistication.

Technology

AP-LIS: Sunquest CoPathPlus 6.0 (Sunquest Information Systems, Tucson, AZ); STS Workstation, 1 Dell Optiplex 745 (Dell, Roundrock, TX, USA) running Microsoft Windows XP; STS™ Lab Volume and Scenario Solutions (Software Testing Solutions™, Tucson, AZ, USA); Remote Connection: PC Anywhere v11.5 (Symantec™, Cupertino, CA, USA).

Design

Working with STS, their software was adapted to validate aspects of the CoPathPlus 6.0 AP-LIS. Specifically, the system will automatically accession cases across all specimen classes inserting all part types and quick text in that specimen class. Further, the system will automatically and iteratively order all available stains, stain protocols, and procedures using a specified number of accessions. If a stain has been previously used in a protocol, it will skip to the next stain in the dictionary. All cases are automatically signed out and transmitted across the HL7 interface. The design leverages all rules system configurations built in the AP-LIS. For example, a quick text associated with a part type will be used preferentially to random quick text. The system can be uniquely configured to test workflow at different sites.

Results

Although the system is very useful for validation of results in a receiving system by generating a large amount of outbound data, a more important benefit is assessment of functionality of AP-LIS. Since the system leverages rules within the AP-LIS, secondary events are generated that can be difficult to test in a manual fashion. All of the cases signed out automatically in STS also generate events such as billing data, slide label, cassette printing, and transactions to interfaced systems such as auto-stainers. Manually, this would take weeks to actualize; as STS is automated, we can perform the same testing for every part type, stain protocol, and procedure with all secondary events in less than 48 h with using the least possible number of accession numbers.

Conclusion

Automation of AP-LIS validation testing can provide significant benefits both in terms of validation testing as well as system functional testing. The solution will continue to be developed and enhanced to be applied to cytopathology testing.

Session-3

Applied Topics

Applied Informatics and Clinical Workflow

Thursday, October 11, 2012 (7:30 am - 8:55 am)


Toronto Room

Mitigating the Risks of Custom Software Development in Pathology using Open Source Software

Peter Gershkovich, John Sinard

Department of Pathology, Yale Medical School, Yale University School of Medicine, New Haven, CT. E-mail: peter.gershkovich@yale.edu

Content

The demands of patient care, financial constraints, and educational challenges produce an increasing pressure to create better technology to handle multiple missions of Pathology Departments. Over the past eight years, the Pathology Informatics group at Yale has developed many applications to handle outreach support, barcoding, and tracking of blocks and slides, frozen section management, residency training tools, document management, monitoring, and notification, etc. This development relied heavily on Open Source software. Using well-tested, production-level, and enterprise-quality common components not only increased the productivity of development but also reduced the amount of custom code. We argue that it in turn mitigates the risk of building and supporting custom software.

Technology

The overwhelming majority of our custom applications and modules have been developed using Java. Additional open source applications, Application Programming Interfaces (APIs), and Frameworks included: Apache HTTPD Server, Apache Tomcat, MySQL database, Quartz scheduling framework, Google Web Toolkit, HAPI API, iText, jFreeChart, Spring, Hibernate, etc.

Design

Most of our applications have a user interface build with Google Web Toolkit. The code is typically deployed on Apache Tomcat following the Servlet 3.0 specification. The middle tier comprises business logic and uses various APIs for common tasks (e.g., iText for PDF rendering, jFreeChart for creation of reports, etc.). An application can have its own relational database management system and/or have access to our core Anatomic Pathology Laboratory Information System database deployed on Sybase ASE

15.03. All custom databases are implemented on MySQL 5.5.

Results

None of the project modules took longer than 6 months from the design to the implementation. Custom development allowed implementation of small functional blocks of code that encapsulated a particular workflow step creating immediate improvements in the process and making it easy to re-engineer the process and implement the entire solution.

Conclusions

The challenges of Pathology Departments at Academic Institutions require development of Information Technology Solutions. Build versus buy approaches have pros and cons. The evolution of Open Source software creates an opportunity for rapid development unparalleled by either of the extremes and reduces the risks of in-house application development.

Implementation of Customer Relationship Management Software to Support Laboratory Outreach at Henry Ford Health System, Henry Ford Medical Laboratories

Mehrvash Haghighi, J. Mark Tuthill, Sarah Mackay

Department of Pathology and Laboratory Medicine, Henry Ford Hospital, Detroit, MI. E-mail: mhaghig1@hfhs.org

Content

To support our laboratory outreach efforts, we implemented and customized Microsoft Dynamicscustomer relationship management (CRM) application to automate and standardize the process of capturing and resolving defects and client issues. This application provides: issue tracking; management and escalation; customer data management, customer configuration data, and sales force support including marketing, call tracking, and request management.

Technology

Microsoft Dynamic CRM 3.0 (3.0.5300.0) purchased through Sunquest Information Systems, Tucson, AZ; Screens and workflow customized by HFHS; Deployed on Windows VMware servers (DL 380 G6, (2) 6-core processors, 24GBRAM, (8) 146 GB drives.)

Design

We modified the design of CRM based on Henry Ford Medical Laboratories (HFML) needs and workflow. We engaged users providing opportunities for decision making, key requirements, current headaches, and preferred work flow for incorporation into our final design. Communications and training strategies were developed to disperse information, and guide users for successful adoption including E-mail announcements, demonstrations, training classes, and quick guides. Current workflow and issues and problems were analyzed by direct observation and parallel testing to allow for effective design.

Results

The implemented CRM module has replaced manual processes that included paper logs, spreadsheets, and manual generation of issues trends data. This consolidated several sources of disparate information from Excel, Word, and MS Access programs into a seamless process. As an integrated, consolidated system, it offers clear visibility and management of customer data.

Conclusions

Moving from a paper-based system to an electronic system has enabled HFML client service to strengthen customer relationships and refine its business processes and work-flow to support greater profitability. Key benefits can be divided into direct benefits such as reduced paper, printing and storage costs and indirect benefits including improved customer service productivity, quality assurance, customer retention, and relationships. CRM has created seamless connectivity and communications between clients, customer service, couriers, and lab personnel through a paperless process automating activities regarding client issues. It also offers a centralized location for sales and marketing notes. By using CRM, employees are better informed regarding the client data and issue history.

The Department of Health and Human Services "Wall of Shame": An Analysis of Large Security Breaches of Protected Health Information

Chiraag D. Patel 1 , Alexis B. Carter 2

1
Department of Pathology, Stony Brook University Medical Center, Stony Brook, NY, 2 Department of Pathology, Emory University School of Medicine, Atlanta, GA. E-mail: chiraag.patel@stonybrook.edu

Content

Health data contain personal information that could be used to harm a patient. Concerns over the ever increasing use of electronic media to manage health information resulted in several federal laws, one of which (Health Information Technology for Economic and Clinical Health Act [HITECH]) requires notification of patients if their health data have been disclosed in an unauthorized manner (security breach). If a security breach of greater than 500 patients occurs, details are made publicly available on the US Department of Health and Human Services (HHS) website. An analysis of this public data set has not been previously published.

Design / Technology

The dataset of all security breaches involving 500 or more patients from inception of HITECH in 2009 was downloaded in March 2012 from the US Department of HHS website. These breaches were analyzed and categorized by year as well as by the type of loss, medium of loss, and number of affected patients.

Results

Over 20,066,249 patients' data were breached, but it was not possible to determine how many of these represented unique patients (maximum of 6.4% of US population affected). Loss of large sets of patient data was primarily due to theft (52.18%) or accidental disclosure (18.39%). The location of the data at the time of loss was overwhelmingly on paper (25.29%) or on a laptop (25.06%). Further detail is present in [Table 1]. The number of reported large security breaches has increased each year, and the most significant increase in overall security breaches occurred from 2009 to 2010 (252.13%).



Conclusion

Recently enacted federal regulations have given the health care industry a much needed push into electronic media to facilitate improvements in patient quality and safety. However, patient harm may occur if sensitive data are accessed or distributed in an unauthorized manner, and health- care entities which must disclose security breaches to patients will incur financial, legal, and reputational difficulties. The analysis of this data set reinforces the security risks inherent with maintaining sensitive data in any format especially data on portable electronic devices. Limiting unwanted security breaches and unauthorized data access should be a high priority for all health care entities.

Automating Grossing in Anatomic Pathology

Shree Sharma 1 , Manisha Singh 2

1
Departments of Pathology, and 2 Internal Medicine, University of Arkansas for Medical Sciences, Little Rock, AR. E-mail: drshreegopal@gmail.com

Content

More than 50% of the specimen load of an anatomic pathology laboratory is constituted by biopsy specimens. The present proposal is related to automation of the anatomic pathology laboratory. The automation will handle the grossing of small specimens not requiring orientation and inking. The current problems in grossing include shortage of staff, mix-up of specimens, and the time devoted to handling of specimens.

Technology

The technology includes redesigning the existing devices which are used in the laboratory on a day-to-day basis.

Design

The design includes a biopsy test tube, cassette, and the robotic system to transfer the specimens. The biopsy-test tube will have an in-built biopsy bag. The biopsy will be collected by the surgeon in this biopsy-test tube with patient identifiers and barcode. The biopsy-test tube after getting accessioned in the pathology lab will be put on the conveyor belts, similar to that used in clinical pathology. A second row of belts with cassettes will move along on a separate belt. The specimen bag will be pulled out of the test tube and dropped in the labeled cassette automatically by the robotic system.

Results

The proposed design and technology will decrease the time needed for processing specimens and will improve the turnaround time. It will eliminate the chances of specimen mix-up and carry over because the specimen will not be opened on the grossing table. It will also reduce the problems related to shortage of staff in the grossing laboratory.

Conclusion

The rapid advancement in the field of digital pathology will change the way we practice pathology in future. The automation of grossing in anatomic pathology is much needed and the lack of attention toward this aspect will act as a bottle neck in the automation and development in the field of anatomic pathology.

Customization of a Digital Pathology Consultation Portal for Clients and Consultants

Somak Roy 1 , William Cable 2 , Eugene Tseytlin 3 , Andrew Lesniak 2 , Jeffery McHugh 2 , Gonzalo Romero Lauro 2 , Samuel Yousem 1 , George Michalopoulos 1 , Anil V. Parwani 1 , Liron Pantanowitz 1

1
Department of Pathology, 2 Information Services Division, 3 Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, PA. E-mail: roys@upmc.edu

Content

Peer review and obtaining second opinions on perplexing cases is an integral part of contemporary anatomical pathology practice. The era of digital pathology facilitates inter-institutional and international pathology teleconsultations. However, a successful telepathology program requires significant customizations to be effective. This study reports our experience with a highly customized web-based portal for handling digital second opinion consultations.

Technology

The portal (https://kingmed.upmc.com) is a web-based interface coded in ColdFusion which enables telepathology by incorporating clinical workflow with a whole slide image (WSI) viewer. The solution leverages a vendor-neutral viewer that is compatible with the most common imaging pathology formats available in the market. This is an internally developed Java applet which supports navigation, annotation, and snapshot functionality in addition to advanced features such as the latest NDPI file format from Hamamatsu. To improve performance, multi-threading capability was added to the viewer.

Design

The portal is based on a client-server model. Using secure login, the client (Kingmed Laboratories, China) scans the slides to an image server and accessions their case with accompanying clinical data into the web-based system. Pathology consultants (UPMC, Pittsburgh) access the WSI on the image server via the web-based collaboration tool and employ the portal to manage workflow (e.g., triage requests), information exchange (e.g., request more history, slides or stains), navigate images (e.g., annotation, snapshot acquisition), for reporting, and transmission of diagnoses to the submitting client.

Results

A total of 33 cases were received for consultation via the portal over 8 months. All the cases were surgical pathology specimens, most commonly breast/female genital tract (n = 10, 30%) and soft tissue (n = 6, 18%). WSI (n = 365, average 11 slides/case) received included 134 H and E, 6 histochemical, and 225 immunohistochemistry slides. Mean turnaround time for 22 consulting pathologists was 40 h (range 2-152 h). Delayed cases were due to network outage problems. Post-launch feedback resulted in further customization to incorporate transcription services, peer-to-peer review for consultants, provision for issuing addenda/amendments and improved WSI viewing experience.

Conclusion

Combining digital imaging with the Internet to create a custom client digital pathology portal broadened our institutions' capability to offer international teleconsultation services. This scalable, secure, and user-friendly web-based collaboration tool permitted cases to be handled via telepathology in a vendor-agnostic manner. Ongoing front- and back-end customization of this solution is responsible for client satisfaction and consultant buy-in.

Automated Review of Laboratory Test Appropriateness

Ronald Hauser

Department of Laboratory Medicine, Yale-New Haven Hospital, New Haven, CT. E-mail: ronald.hauser@yale.edu

Content

Inappropriate use of laboratory services produces billions of dollars in medical waste annually. The determination of the appropriateness of a laboratory request involves a cost-benefit analysis by placing relative utility on the test cost, iatrogenic injury in specimen procurement, the potential for misleading results, and favorable changes to patient management prompted by test outcomes. Weighing these factors results in a threshold that varies with pre-test probability. In everyday practice, clinicians use the patient presentation to assess the pre-test probability and recommend lab tests. Quantification of this intuitive process forms a barrier to automation. This paper proposes an alternative approach using Bayes' Law to retrospectively calculate pre-test probability using test outcomes.

Technology

The application of this approach to laboratory information systems could systematically identify low-yield tests for utilization review.

Design

The study compared the pre-test probability of West Nile meningoencephalitis in a patient population referred for West Nile IgM from cerebrospinal fluid using clinical features and Bayes' Law. A literature review identified clinical signs with high sensitivity for West Nile meningoencephalitis: seasonality (August, September), age (≥50), fever (≥100.4°F before or during evaluation), and CSF pleocytosis (≥5/mm 3 ). A retrospective chart review of 439 test requests recorded these signs. To obtain the inputs for the Bayes' calculation, test results were obtained. The product insert provided estimates of sensitivity (95%) and specificity (99%).

Results

Five of the 439 tests had a positive result. The Bayesian pre-test probability had a value <1% [(0.99 + 5/439-1)/(0.99 + 0.95-1)]. The clinical assessment of pre-test probability found 11 of 439 patients (2.5%) meeting all four criteria for West Nile meningoencephalitis. Of these 11 cases, 4 had a positive result (36%). The single case not satisfying all four features missed only the age criteria by 1 year.

Conclusion

Pre-test probability calculated with Bayes' law and a clinical assessment both predicted a low disease prevalence in the population referred for testing. The application of Bayes' laws allows laboratorians to determine the threshold used by clinicians to place requests on laboratory services. It permits the systematic identification of low-yield diagnostic testing, which should prompt a utilization review. Automated determination of pre-test probability increases its feasibility as a pay-for-performance quality metric [Figure 1].



Session-4

Decision Support

Applied Informatics and Clinical Workflow

Thursday, October 11, 2012 (7:30 am - 8:55 am)


New Orleans Room

Real-Time Detection and Display of Critical Values in the Laboratory Improves Caregiver Notifications

Kavous Roumina, Eugene Farber, Walter H. Henricks

Center for Pathology Informatics, Cleveland Clinic, Cleveland, OH.

E-mail: roumink@ccf.org

Content

Timely notification to caregivers of critical test results requiring immediate attention is an important function of the laboratory and necessary for accreditation compliance. In high volume laboratories, improvements in alerting laboratory staff to the presence of unreported critical values could improve time to communication of such results and reduce call back failures. We developed an automated system that detects and displays to laboratory staff the pending critical results not yet communicated to caregivers.

Technology

Data analysis and presentation components (.NET Framework 3.5, Internet Information Services, Microsoft); relational database (MS SQL Server 2005, Microsoft); laboratory information system (LIS) (Sunquest Laboratory).

Design

The system consists of controller and presentation modules, both of which use the same back-end database. The controller continually queries the LIS, storing in the database only results not denoted as "called." The web-based presentation module displays on large monitors color-coded critical results, yellow when first displayed, red within 30 min of availability from the analyzer. After caregiver notification, a technologist adds a predefined result modifier in the LIS allowing for the result removal from the display. Easily configurable system parameters enable inclusion of additional test types.

Results

The system processed over 3.5 million results from the LIS (potassium and glucose; February-July, 2012). On average, 835 results/h were analyzed (24 h cycle), of which 1.7 results/h were displayed as critical results. For critical outpatient results, where timely communication is more difficult than for an inpatient, the laboratory met monthly targets for percentage results called back (92% within 30 min) in five of 6 months (83%) following implementation, compared with only three of 6 months (50%) in the same months in 2011. For inpatient critical results, the laboratory met targets in all 6 months (100%) compared to five of 6 months (83%) in 2011.

Conclusion

We developed an automated system that detects critical results not yet communicated to caregivers and displays them efficiently to laboratory staff through a web browser. Following implementation, the laboratory realized improvements in meeting targets for critical value reporting time. The design paradigm may be extended to "urgent" results as well.

Utilization of Clinical Laboratory Reports with Graphical Elements

Brian H. Shirts 1 , Nichole Larsen 2 , Brian R. Jackson 2,3

1
Department of Laboratory Medicine, University of Washington, Seattle, WA, 2 ARUP Laboratories, Salt Lake City, UT, 3 Department of Pathology, University of Utah, Salt Lake City, UT. E-mail: shirtsb@uw.edu

Content

Graphical reports that contain charts, images, and tables have the potential to convey information more effectively than text-based reports; however, studies have not measured how much clinicians value such features. We sought to identify factors that might influence the utilization of reports with graphical elements postulating that this is a surrogate for relative clinical utility of these graphical elements.

Technology

ARUP Laboratories established pilot projects to test an online enhanced electronic laboratory reporting system in November 2009 to provide clinicians with graphical elements that cannot be transmitted or displayed with current interface systems. A URL along with chart id number and password are provided as a footnote to the LIS-version test result. When a test result is verified within the laboratory information system, a pdf-format chart is generated and stored on a server can only be accessed by those that have access to the password transmitted with reports via current interfaces. Login information linked to specific test reports allowed us to track how often these reports were accessed.

Design

We evaluated utilization of enhanced electronic laboratory reports from institutions across the United States. We monitored on-demand clinician access to reports generated for 48 reportable tests over 22 months. We evaluated utilization of 174,170 reports with graphical elements by clinicians at all institutions that use ARUP as a reference laboratory. We used descriptive statistics, regression, and meta-analysis tools to evaluate groups of similar test reports. We compared download rates for tests with different forms of graphical and photographic information to several reports which were generated with standard formatting but no additional clinical information.

Results

Median download rate by test was 8.6% with high heterogeneity in download rates between tests. Test reports with additional graphical elements were not necessarily downloaded more often than reports without these elements. Recently implemented tests and tests reporting abnormal results were associated with higher download rates (P < 0.01). Higher volume tests were associated with lower download rates (P = 0.03).

Conclusion

In select cases, graphical information may be clinically useful, particularly for less frequently ordered tests and in reports of abnormal results; however, between-test heterogeneity was high. The utilization data presented could be used as a reference point for other laboratories planning on implementing graphical reporting. In many cases graphical elements may add little clinical utility, particularly if these merely reinforce information already contained in text-based reports.

Computerized Provider/Physician Order Entry in Anatomic Pathology: A Single Center Experience

Seung Park 1 , Anil Parwani 1 , Samuel Yousem 1 , Luke Wiehagen 1 , Susan Kelly 1 , Tony Piccoli 2 , Frank J. Losos 2 , Kara Balatincz 1 , Liron Pantanowitz 1

1
Department of Pathology, 2 Information Services Division, University of Pittsburgh Medical Center, Pittsburgh, PA. E-mail: parks3@upmc.edu

Content

Although computerized provider/physician order entry (CPOE) modules in the electronic medical record (EMR) and laboratory information system (LIS) have become common in clinical pathology, their adoption in anatomic pathology (AP) has been slower. The aim of this study was to present our experience with a novel implementation of AP CPOE in a large academic hospital.

Technology

APLIS: CoPathPlus (version 3.2, Cerner); EMR: SurgiNet (Cerner Millenium).

Design

Respective modules in the APLIS and EMR were customized in collaboration with the vendor, enabling bidirectional communication of surgical order data to accompany surgical pathology specimens at one hospital (UPMC Shadyside). Surgical nurses entered surgery type and specimen descriptions with orders into the EMR. This order information, along with additional clinical data, was extracted into an APLIS-bound HL7 message. This message was used to accession specimens and populate case information in the APLIS. Adverse event data regarding errors in specimen accessioning were compared for specimens received with and without CPOE. Nurses were then trained in CPOE, and a second round of data collection ensued.

Results

A total of 17,924 surgical cases were performed at UPMC Shadyside in a 6-month interval, 4403 (25%) of which utilized CPOE. There were 92 adverse events reported among CPOE-enabled surgeries (2%) and 127 adverse events among non-CPOE-enabled surgeries (1%). Adverse events due to missing/incorrect patient information, clinical history/procedure, and clinician name or other registration issues were completely eliminated in CPOE-enabled surgeries. However, adverse events related to missing/incorrect specimen information, discrepancy between specimen container and paper requisition, and truncation of data received due to inappropriate character usage showed sharp increases, together accounting for the vast majority of adverse events reported in CPOE-enabled surgeries. When there was AP CPOE training, adverse events dropped by 75% [Figure 1].

Conclusion

AP CPOE is feasible and has benefits similar to CPOE in clinical pathology. The increase in adverse events reported in CPOE-enabled surgical cases is largely attributed to the fact that surgical nurses are not adequately trained to properly accession specimens. In conclusion, although AP CPOE is novel at present, further development in this field is anticipated, as the benefits clearly outweigh the challenges.



Impacting Patient Care through Laboratory Decision Support-Identification and Alert of the Patient's First Troponin-I Increase

Eugenio Zabaleta 1 , John Burgess 1 , Gregory Eaton 2 , Michael Patterson 3

1 Department Laboratory, and 2 Cardiology, 3 Administration, MedCentral Health System, Mansfield, OH. E-mail: ezabaleta@medcentral.org

Content

MedCentral Physicians requested the laboratory to identify clinically significant changes of Cardiac Troponin results and alert those findings to the caregivers in a timely fashion in order to impact patient care. Cardiac troponin (cTn) is the recommended biomarker for the detection of myocardial injury or necrosis when assessing acute myocardial infraction by the American College of Cardiology. Information technology (IT) can be used to improve communication for better patient outcomes.

Technology

To address the physicians' request, MedCentral has incorporated a Laboratory Decision Support Software called RippleDown® (Pacific Knowledge Systems™). Within this software, rules were created for the purpose of providing clinically relevant alerts based on the outcome of current and previous cTn testing for each patient to warn physicians of potential clinical issues with their inpatients. Consistent application of clinical knowledge through this software enhances communications between the hospital clinical services and has the possibility of decreasing human errors.

Design

Since March 2010, MedCentral Laboratory has been able to identify clinically significant changes of Cardiac Troponin using this artificial intelligence. When the first increase of troponin-I occurs, RippleDown® will send an alert message to the lab information system. The performing lab professional will call the patient's nurse and communicate the first increase of troponin-I. The nurse will then phone the physician. The results will also be communicated to the attending and/or consulting physicians through the Hospital Information System under the Alert section through Rules and Workflow software engine.

Results

MedCentral Laboratory compared pre- and post-implementation data to study the impact of this project. The control group is from December 2009 to February 2010 (pre-implementation); during this period 5195 cTn were performed, 151 would be identified as first cTn increases. The study group was from December 2010 to February 2011; during this period 4755 cTn were performed, 140 were identified as first cTn increases. The most clinically important finding was a significant reduction of length of stay in patients admitted in non-cardiac services with non-cardiac related admitting diagnosis from 7.89 to 6.17 days.

Conclusion

First increased troponin-I alert project is improving timely communication between hospital's services that leads to better and more effective patient care.

Use of Automated Algorithms to Detect Early Acute Kidney Injury

Jason M. Baron 1 , Xingxing S. Cheng 2 , Hasan Bazari 2 , Jingyang Huang 1 , Rosemary T. Jaromin 1 , Kent B. Lewandrowski 1 , Anand S. Dighe 1

1
Departments of Pathology, 2 Medicine, Massachusetts General Hospital, Boston, MA. E-mail: jmbaron@partners.org

Content

Acute kidney injury (AKI) is a common problem in hospitalized patients. Clinical criteria (e.g., RIFLE) allow a diagnosis of AKI to be made based on specified increases in blood creatinine. However, clinical laboratories only flag creatinine results exceeding the reference range and generally do not flag creatinine increases from an individual patient's baseline. Consequently, AKI may be overlooked in patients with rising creatinine values that still fall within the reference range. In this report, we evaluate rule-based strategies, implementable within our laboratory information system that could be used to flag rising creatinine levels and facilitate early identification of AKI.

Technology

Data were extracted from our hospital's laboratory information system and analyzed using relational database tools.

Design

We evaluated three different types of flagging systems. The first was a delta check, comparing the current creatinine value to the patient's most recent value within the prior 72 h. The second was based on comparing the patient's current creatinine value to a 72 h, tracked-minimum value. The third was a combination of a delta check and a tracked-minimum-based comparison. Rules were compared to RIFLE criteria using 42,299 inpatient creatinine results from our hospital.

Results

After optimization of the specific creatinine change thresholds, rules combining delta checks with tracked minimum comparisons generally demonstrated improved performance with respect to RIFLE criteria in comparison to rules based on delta checks alone or on tracked minimum values alone. For example, one combination rule was more than 80% sensitive and 80% specific in identifying AKI (injury level criteria). Moreover, we found that certain rules may be useful in identifying patients at risk for proceeding to AKI before they meet formal criteria.

Conclusion

Rule-based flagging systems within a laboratory information system may be useful in identifying patients at risk for AKI.




 

 
Top
  

    

 
  Search
 
   Browse articles
  
    Similar in PUBMED
    Access Statistics
    Email Alert *
    Add to My List *
* Registration required (free)  

 
  In this article
    Electronic Poste...
   Scientific Sessions
    Electronic Poste...
   Scientific Session

 Article Access Statistics
    Viewed9032    
    Printed161    
    Emailed0    
    PDF Downloaded596    
    Comments [Add]    

Recommend this journal