PROJECT TITLE :

Lung cancer survival prediction from pathological images and Genetic data - an integration study - 2016

ABSTRACT:

In this paper, we tend to have proposed a framework for lung cancer survival prediction by integrating genetic data and pathological images. Since molecular profiles and pathological images reveal complementary information on tumor characteristics, the combination can benefit the survival analysis. The gene expression signatures are processed using Model-Primarily based Background Correction methodology. A sturdy cell detection and segmentation methodology is applied to phase every individual cell from pathological pictures to extract the image options. Based mostly on the cell detection results, a set of extensive features are extracted using economical geometry and texture descriptors. The supervised principal part regression model is fitted to evaluate the proposed framework. Experimental results demonstrate strong prediction power of the statistical model built from the combination of genetic data and pathological pictures compared with using only one of the 2 types of data alone.