Chao Chen (Stony Brook)
Topology-Informed Image Segmentation, Generation and Analysis
Thanks to decades of technology development, we are now able to visualize in high resolution complex biomedical structures such as neurons, vessels, trabeculae and breast tissues. We need innovative approaches to fully exploit these structures, which encode important information about underlying biological mechanisms. In this talk, we explain how topological information can be seamlessly incorporated into different parts of a deep learning pipeline, based on the theory of persistent homology. This leads to a series of novel methods for better segmentation, generation, and analysis of these topology-rich biomedical structures. We will also briefly mention a recent work using persistent homology to analyze the behavior of neural networks under backdoor attacks.