Aorta Registration and Deformation Analysis

CT-based deformable image registration methods for measuring and characterizing thoracic aortic aneurysm growth, with clinical validation.

This project develops deformable image registration techniques for quantifying thoracic aortic aneurysm (TAA) growth from longitudinal CT scans, conducted with Prof. Nicholas Burris at the University of Michigan Department of Radiology.

Key contributions:

  • Vascular Deformation Mapping (VDM) validation: Created 76 synthetic TAA growth phantoms via 3D Blender to validate the VDM pipeline, demonstrating that VDM’s precision significantly outperforms traditional diameter assessments by human raters. Published in Medical Physics (2022).

  • Clinical translation: VDM was applied to CT surveillance of thoracic aortic aneurysm growth, published in Radiology (IF: 29.1, 2022), with 40+ citations.

  • Directional growth analysis: Developed a technique to decouple TAA growth into longitudinal and radial components, validated on both digital phantoms and real-patient CT datasets. Published at SPIE Medical Imaging 2021.

  • LitCall: Designed a deep learning method for aortic landmark detection that incorporates geometric priors, achieving superior localization accuracy with minimal model size increase. Published at SPIE Medical Imaging 2022.

  • Automated aortic measurements: Contributed to a fully automated pipeline using joint segmentation and localization neural networks. Published in Journal of Medical Imaging (2023).

References