Shape Characterization of Extracted and Simulated Tumor Samples using Topological and Geometric Measures

Abstract

The prognosis of cancer patients suffering from solid tumors significantly depends on the developmental stage of the tumor. For cervix carcinoma the prognosis is better for compact shapes than for diffusive shapes since the latter may already indicate invasion, the stage in tumor progression that precedes the formation of metastases. In this paper, we present methods for describing and evaluating tumor objects and their surfaces based on topological and geometric properties. For geometry, statistics of the binary object's distance transform are used to evaluate the tumor's invasion front. In addition, a simple compactness measure is adapted to 3D images and presented to compare different types of tumor samples. As a topological measure, the Betti numbers are calculated of voxelized tumor objects based on a medial axis transform. We further illustrate how these geometric and topological properties can be used for a quantitative comparison of histological material and single-cell-based tumor growth simulations.