|J Pathol Inform 2017,
Commentary: Improving the efficiency of the ova and parasite examination using cloud-based image analysis
Daniel D Rhoads
Department of Pathology, Case Western Reserve University; University Hospitals Cleveland Medical Center, Cleveland, OH, USA
|Date of Submission||23-Sep-2017|
|Date of Acceptance||03-Nov-2017|
|Date of Web Publication||14-Dec-2017|
Dr. Daniel D Rhoads
University Hospitals Cleveland Medical Center, 11100 Euclid Avenue, Mailstop: PTH 5077, Cleveland, Ohio 44106
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Rhoads DD. Commentary: Improving the efficiency of the ova and parasite examination using cloud-based image analysis. J Pathol Inform 2017;8:49
| Summary|| |
In Holmström et al.'s proof-of-concept study, “Point-of-care mobile digital microscopy and deep learning for the detection of soil-transmitted helminths and Schistosoma haematobium,” the authors use two emerging technologies that have the potential to shake up the routine ova and parasite (O&P) examination. First, the authors used a portable digital microscope that costs “approximately that of a midrange smartphone” to capture images of a quality comparable to using a × 10 objective lens on a traditional microscope. Many studies have investigated the utility of rugged, portable digital microscopes, but this system was different from most in that it has motorized movement capability, which works to capture numerous representative fields from a single slide. Second, the authors trained a commercially available cloud-based image analysis software, WebMicroscope (https://webmicroscope.com/imageanalysis/), to perform a two-step analysis pipeline. The first algorithm in the pipeline identified objects of interest (i.e., rare events) that were suspected to be parasite eggs. The second algorithm classified the identified objects based on morphology as either an artifact or an Ascaris egg, a Trichuris egg, or a hookworm egg.
In the test set used to challenge the O&Panalysis pipeline, the large majority (90%) of eggs analyzed were from Ascaris. One hundred and ninety-five Ascaris eggs were present, and the software identified all of them as such with only 13 false-positive events. The system was also able to detect and discriminate Trichuris and hookworm eggs in stool. Schistosoma eggs were able to be identified manually in urine using the microscope system, but the article does not describe the development of software to detect the Schistosoma eggs. Notably, the study was designed to focus on the sensitivity of the software over the specificity of the software as images without eggs were not analyzed, so the potential false-positivity rate if the pipeline were to be used in clinical practice is unclear.
| Comments|| |
Image analysis software is on the verge of becoming mainstream in clinical microbiology as a growing number of laboratories are converting to digital plate reading of bacterial cultures as a component of total laboratory automation. Laboratories and manufacturers are exploring ways to add image analysis software to the digital plate reading process to further improve efficiencies., Another analysis platform that has been used in infectious disease diagnosis is CellaVision, which was designed for assisting with manual white blood cell differentials. CellaVision has been used successfully to detect malarial parasites in digitized peripheral blood smears. In Holmström et al.'s discussion of their study of O&P examinations, they suggest that their hardware and software solution could be further developed to facilitate O&P examinations in resource-limited environments, which is true. However, further developing and clinically validating a Holmström system to be used in the routine O&P examination workflow in resource-rich settings would also be appropriate.
This proof-of-concept study reveals an opportunity to improve the efficiency of clinical microbiology practice. Implementing a clinical testing system similar to what is described by Holmström et al. would be a leap forward in clinical parasitology in the United States. The Holmström study trained an off-the-shelf, cloud-based image analysis tool that was developed for general pathology to successfully identify and classify ova from parasites in a stool specimen. Having this software available in the cloud facilitates sharing of the algorithm with other laboratories that could potentially validate and use the software for clinical testing purposes. A resource that is needed, which could help to expedite the development of effective artificially intelligent O&P image analysis tools, is the formal curation of a freely available image library comprised of well-annotated images that developers can use to train and test software classifiers, similar to the MicrobIA project for bacterial cultures (www.microbia.org).
If an image analysis solution could be validated for clinical use in O&P examinations, it could do to the O&P examination what CellaVision has done to the manual white blood cell differential. When implemented, the Holmström system would present a number of candidate images to a technologist who could quickly visually analyze the images and determine whether the software's classification of the objects as stool artifacts or stool pathogens was correct (https://goo.gl/rKrbge). Clinical validation and laboratory implementation of an O&Psolution that is similar to what Holmström et al. have described would be a huge advance for clinical parasitology. The system would have the potential to decrease the analysis time of the O&P examination from 8 to 10 min per specimen to just seconds.
| References|| |
Holmström O, Linder N, Ngasala B, Mårtensson A, Linder E, Lundin M, et al.
Point-of-care mobile digital microscopy and deep learning for the detection of soil-transmitted helminths and Schistosoma haematobium
. Glob Health Action 2017;10:1337325.
Holmström O, Linder N, Lundin M, Moilanen H, Suutala A, Turkki R, et al.
Quantification of estrogen receptor-alpha expression in human breast carcinomas with a miniaturized, low-cost digital microscope: A Comparison with a high-end whole slide-scanner. PLoS One 2015;10:e0144688.
Rhoads DD, Novak SM, Pantanowitz L. A review of the current state of digital plate reading of cultures in clinical microbiology. J Pathol Inform 2015;6:23.
] [Full text]
Faron ML, Buchan BW, Coon C, Liebregts T, van Bree A, Jansz AR, et al.
Automatic digital analysis of chromogenic media for vancomycin-resistant-enterococcus screens using Copan WASPLab. J Clin Microbiol 2016;54:2464-9.
Ferrari A, SLombardi S, Signoroni A. Bacterial colony counting with convolutional neural networks in digital microbiology imaging. Pattern Recognit 2017;61:629-40.
Racsa LD, Gander RM, Southern PM, McElvania TeKippe E, Doern C, Luu HS, et al.
Detection of intracellular parasites by use of the CellaVision DM96 analyzer during routine screening of peripheral blood smears. J Clin Microbiol 2015;53:167-71.
Kratz A, Bengtsson HI, Casey JE, Keefe JM, Beatrice GH, Grzybek DY, et al.
Performance evaluation of the CellaVision DM96 system: WBC differentials by automated digital image analysis supported by an artificial neural network. Am J Clin Pathol 2005;124:770-81.