For the development of CAD systems of prostate cancer, manual annotation of cancer by experienced pathologists is the gold standard for establishing the ground truth. However, the process is tedious and has finite precision. Here, we describe a framework that uses quantitative analysis of IHC-stained slides to derive parameters, which in turn are used by a trained predictive model to estimate the spatial distribution of malignant epithelium. Thresholding of the results provides a reasonable map of cancer that is comparable to manual annotation.
This abstract and the presentation materials are available to members only; a login is required.