Computational Analysis of 3d Cleared and Labeled Pancreatic Cancer Samples

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is the most common form of pan- creatic cancer and with a 5-year survival rate of only 11%, it is amongst the deadliest cancers. Understanding pancreatic cancer growth pattern on a microscopic scale in 3D will help to understand mechanisms of metastasis and invasion. Modern Tissue Clearing can be used to obtain microscopic fluorescent 3D images of real patient tissue. In this thesis, we build the pipeline for analyzing PDAC structures in 3D using fluorescent microscopic images by generating surface models and quan-tifying them by centerline, persistent homology, endpoints, and sphericity analysis.