Heart Structure Segmentation

Cardiac CT Angiography is commonly employed for cardiac diagnosis. Quantification and extraction of the heart structures are used for disease diagnosis and treatment planning. This project automatically segmented the four chambers of the heart, myocardium, and the aorta based on deep learning techniques applied to cardiac CTA studies.

This tool uses an advanced deep learning model built from a large-scale annotated dataset to segment four chambers of the heart, including left ventricle, right ventricle, left atrium, and right atrium from CTA. The myocardium of the left ventricle and the ascending aorta are also automatically extracted. Quantifications of those segmentations can be used to assist in the diagnosis of cardiovascular diseases and treatment planning.

  • 3D segmentation of the four heart chambers, myocardium, and ascending aorta
  • Interactive 3D visualization of the segmented structures are overlaid with original DICOM images