image: Learning an image-based model to predict survival. Processed images from patients alive at five years after surgery and from patients deceased at five years after surgery were used to construct an image-based prognostic model. After construction of the model, it was applied to a test set of breast cancer images (not used in model building) to classify patients as high or low risk of death by 5 years. After training the computational predictive model, the authors then applied it to new breast cancer cases and evaluated whether the computer could predict a patient's survival based on analysis of cancer microscopic images. In the study, the predictive model was a strong predictor of patient survival in breast cancer. In addition, the authors analyzed the predictive model to identify new breast cancer morphologic features that are predictive of patient survival. An unexpected finding from the analysis was that the most predictive features were not from the cancer cells themselves, but from the cancer stroma. This is image 1 (D) from the paper. This image relates to a paper that appeared in the Nov. 9, 2011, issue of Science Translational Medicine, published by AAAS. The paper, by Dr. A.H. Beck of Stanford University School of Medicine in Stanford, Calif., and colleagues, was titled, "Systematic Analysis of Breast Cancer Morphology Uncovers Stromal Features Associated with Survival." view more
Credit: Image courtesy of <i>Science</i>/AAAS