From whole slide tissues to knowledge: Mapping sub-cellular morphology of cancer

Abstract

Digital pathology has made great strides in the past decade to create the ability to computationally extract rich information about cancer morphology with traditional image analysis and deep learning. High-resolution whole slide images of cancer tissue samples can be analyzed to quantitatively extract and characterize cellular and sub-cellular phenotypic imaging features. These features combined with genomics and clinical data can be used to advance our understanding of cancer and provide opportunities to the discovery, design, and evaluation of new treatment strategies. Researchers need reliable and efficient image analysis algorithms and software tools that can support indexing, query, and exploration of vast quantities of image analysis data in order to maximize the full potential of digital pathology in cancer research. In this paper we present a brief overview of recent work done by our group, as well as others, in tissue image analysis and digital pathology software systems.