Development of an anatomical atlas for stroke classification with Electrical Impedance Imaging for Deep Learning
Title : Creating Development of an anatomical atlas for stroke classification with Electrical Impedance Imaging for Deep Learning Mentor : Dr. Sarah Hamilton
Electrical Impedance Tomography (EIT) uses harmless electrical measurements taken at the surface of a boundary to recover (or `see') the internal conductivity. Since conductivity is tissue dependent, there are a wide range of applications including heart and lung imaging, breast cancer detection, and stroke classification. This project focuses on developing an anatomical atlas from a set of CT and MR scans of human brains. Segmentation of the CT/MR scans can be used to extract boundaries of the internal structures (head shape, skull location and thickness, spinal cord, etc.) for both healthy and sick patients. The atlas can then be used to generate training data (by solving a partial differential equation with the internal conductivity defined by the anatomical atlas) in a deep learning scheme such as a computational neural network with a U-net structure.