Contact us
|
Home
|
Login
| Users Online: 424
Feedback
Subscribe
Advertise
Search
Advanced Search
Month wise articles
Figures next to the month indicate the number of articles in that month
2022
January
[
3
]
2021
November
[
2
]
September
[
3
]
August
[
1
]
June
[
2
]
January
[
1
]
2020
November
[
3
]
August
[
1
]
July
[
1
]
May
[
1
]
February
[
1
]
2019
December
[
2
]
September
[
1
]
August
[
2
]
July
[
2
]
June
[
1
]
May
[
1
]
April
[
1
]
March
[
1
]
February
[
2
]
2018
December
[
4
]
November
[
1
]
August
[
1
]
July
[
1
]
May
[
1
]
2017
October
[
1
]
September
[
3
]
June
[
1
]
May
[
1
]
March
[
1
]
February
[
1
]
2016
April
[
1
]
March
[
1
]
January
[
2
]
2015
October
[
3
]
September
[
3
]
June
[
4
]
March
[
2
]
January
[
1
]
2014
October
[
2
]
September
[
2
]
August
[
2
]
July
[
1
]
June
[
1
]
May
[
1
]
March
[
1
]
January
[
2
]
2013
December
[
2
]
November
[
1
]
July
[
1
]
June
[
1
]
March
[
2
]
2012
December
[
1
]
September
[
3
]
August
[
1
]
July
[
1
]
April
[
3
]
March
[
1
]
February
[
1
]
2011
August
[
2
]
July
[
2
]
June
[
1
]
May
[
1
]
March
[
2
]
January
[
1
]
2010
October
[
3
]
» Articles published in the past year
To view other articles click corresponding year from the navigation links on the left side.
All
|
Abstracts
|
Book Review
|
Commentary
|
Editorial
|
Letters to Editor
|
Original Article
|
Original Articles
|
PV16 Abstracts
|
Research Article
|
Review Articles
|
Symposium
|
Technical Note
Export selected to
Endnote
Reference Manager
Procite
Medlars Format
RefWorks Format
BibTex Format
Show all abstracts
Show selected abstracts
Export selected to
Add to my list
Research Article:
Training nuclei detection algorithms with simple annotations
Henning Kost, André Homeyer, Jesper Molin, Claes Lundström, Horst Karl Hahn
J Pathol Inform
2017, 8:21 (15 May 2017)
DOI
:10.4103/jpi.jpi_3_17
PMID
:28584683
Background:
Generating good training datasets is essential for machine learning-based nuclei detection methods. However, creating exhaustive nuclei contour annotations, to derive optimal training data from, is often infeasible.
Methods:
We compared different approaches for training nuclei detection methods solely based on nucleus center markers. Such markers contain less accurate information, especially with regard to nuclear boundaries, but can be produced much easier and in greater quantities. The approaches use different automated sample extraction methods to derive image positions and class labels from nucleus center markers. In addition, the approaches use different automated sample selection methods to improve the detection quality of the classification algorithm and reduce the run time of the training process. We evaluated the approaches based on a previously published generic nuclei detection algorithm and a set of Ki-67-stained breast cancer images.
Results:
A Voronoi tessellation-based sample extraction method produced the best performing training sets. However, subsampling of the extracted training samples was crucial. Even simple class balancing improved the detection quality considerably. The incorporation of active learning led to a further increase in detection quality.
Conclusions:
With appropriate sample extraction and selection methods, nuclei detection algorithms trained on the basis of simple center marker annotations can produce comparable quality to algorithms trained on conventionally created training sets.
[ABSTRACT]
[HTML Full text]
[PDF]
[Mobile Full text]
[EPub]
[Citations (2) ]
[PubMed]
[Sword Plugin for Repository]
Beta
Sitemap
|
What's New
Feedback
|
Copyright and Disclaimer
|
Privacy Notice
© Journal of Pathology Informatics | Published by Wolters Kluwer -
Medknow
Online since 10
th
March, 2010