HomeWhy Challenges?All ChallengesCreate Your Own ChallengeContributorsForum
Sign in / Register

 

Team: Smart Imaging Technologies Co., US

Authors: Vitali Khvatkov, Alexei Vylegzhanin

Abstract:

We used two-step multiscale analysis with ensemble of statistical learning models trained on structure and phase descriptors from the image patches. First step is selection of candidates at magnification of 4 microns per pixel. At this step, we used texture descriptor (27 features) with ensemble of Support Vector Machine Models to select image patches for further analysis. Second step is analysis of candidates with boosted multiscale model cascade. We separated Hematoxylin and Eosin components of the image and extracted structural and phase descriptors for those components at magnifications of 1 micron per pixel and 0.5 micron per pixel. We have used combination of these descriptors to train boosted cascade of statistical learning models as classifier.

Results:

The following figure shows the receiver operating characteristic (ROC) curve of the method.

 

The following figure shows the free-response receiver operating characteristic (FROC) curve of the method.

 

The table below presents the average sensitivity of the developed system at 6 predefined false positive rates: 1/4, 1/2, 1, 2, 4, and 8 FPs per whole slide image.

FPs/WSI0.250.51248
Sensitivity0.1600.1690.1910.2220.2400.267


Consortium for Open Medical Image Computing © 2012-