Peter Lawson Seminar - Investigating Topological Descriptors for the Grading of Prostate Cancer
- Tuesday, June 6, 2017 from 11:00am to 12:00pm
- Barnard Hall - view map
Investigating Topological Descriptors for the Grading of Prostate Cancer
Diagnosis of prostate cancer is achieved by evaluating architectural characteristics of biopsy tissue and grading cancer aggressiveness on these features. The current standard for prostate cancer grading is the modified Gleason score, which is a subjective rating system based on an analysis of high-level tissue architecture and glandular shape and organization. Gleason scoring exhibits high intra-observer and inter-observer variability between pathologists, and may not capture all potentially prognostic information contained in glandular growth patterns. We applied persistent homology with the goal of capturing architectural features of prostate glands in pathology images across a range of scales. In this work, we demonstrate the potential effectiveness of persistent homology to recognize and quantify glandular architectural features in a quantitative and reproducible way. Our results point to the ability to quantify distributions of gland cycles and cycle inter-relatedness, with the further ability to window the diagrams and isolate features contributing to the diagrams, which is useful in interpreting quantitative shape descriptors in a histological context. Ongoing work is aimed at investigating classification of an image library based on these descriptors with comparisons to standard Gleason grade classifications.