The document discusses the application of topological data analysis (TDA) in medical image analysis, specifically focusing on understanding transcriptomic characteristics in autism spectrum disorder (ASD) and colorectal cancer (CRC). It outlines the use of persistent homology to analyze high-dimensional gene expression data and proposes a CNN-based method for fast tumor-region segmentation in CRC using features derived from persistent homology. Results indicate that ASD patients exhibit more heterogeneous gene expression profiles compared to controls, and the proposed segmentation method shows improved classification performance over existing techniques.