New in situ data for the analysis of interactions between cancer and the immune system in auxillary lymph nodes
Wednesday October 24, 2007
04:15PM - 05:00PM
Speakers:
Susan Holmes
Abstract:
Recent work (Kohrt et al, 2005) has shown that T cell profiles in lymph nodes are much more accurate predictors of cancer survival in breast cancer than tumor size. In order to extend this preliminary staudy on 76 patients to a very large cohort, we needed to collect large amount of data on stained images. We have built {GemIdent: http://www.gemident.com}} a novel object identification algorithm in Java to locate immune and cancer cells in images of immunohistochemically-stained lymph node tissue from the Kohrt study and also shows promise in other domains. Our method leans heavily on the use of color and the relative homogeneity of object appearance. As is often the case in segmentation, an algorithm specifically tailored to the application works better than using broader methods that work passably well on any problem. Our main innovation is interactive feature extraction from color images. We also enable the user to improve the classification with an interactive visualization system. This is then oupled with the statistical learning algorithms and intensive interaction with the user over many classification-correction iterations, resulting in a highly accurate and user-friendly solution. At the web site we have made available both a detailed manual, movies showing the various stages of analyses and the code as well as the documented source code available to academics.
Kohrt, H.~E., N.~Nouri, K.~Nowels, D.~Johnson, S.~Holmes, and P.~P. Lee, 2005:
Profile of immune cells in axillary lymph nodes predicts disease-free
survival in breast cancer. {\em PLoS Med\/}, {\bf 2(9)}, e284.
Kapelner, A., P.~Lee, and S.~Holmes, 2007: An interactive statistical image segmentation and visualization system. {\em Medivis\/}, {\bf 00}, 81--86.
|
