Image Segmentation for 3D Heart Models
In this project I will develop image segmentation methods to delineate the heart in 3-D cardiac MRI images of patients with complex congenital heart disease. Image segmentation involves labeling each voxel with the region of the heart that it lies in. Segmenting cardiac images is important for creating patient-specific 3-D models of the heart which can be displayed virtually or 3-D-printed to help clinicians plan surgery for congenital heart disease. Currently patient-specific 3-D heart models are underused because it takes around 4-8 hours to manually segment cardiac MRI images since each contains approximately 1503 voxels. The dataset I will be working with is from Boston Children’s Hospital which contains the 3-D MRI images and corresponding manual segmentations.

Through this SuperUROP I want to gain more experience in medical image analysis research as well as make a positive contribution to the group. I’ve taken machine learning courses and I want to expand on that knowledge with real world applications. I hope to publish a paper by the end of the SuperUROP if I have meaningful results to display.