CNN-Based Segmentation of the Cardiac Chambers and Great Vessels in Non-Contrast-Enhanced Cardiac CT

TL;DR

This paper introduces a CNN-based method for automatic segmentation of cardiac chambers and vessels in non-contrast cardiac CT, using virtual non-contrast images and aligned CCTA references to improve cardiovascular risk assessment.

Contribution

The study presents a novel training approach for FCNs using VNC images and aligned CCTA data, addressing generalization issues in non-contrast CT segmentation.

Findings

Effective segmentation of cardiac structures demonstrated

Improved accuracy over traditional methods

Potential for enhanced cardiovascular risk stratification

Abstract

Quantification of cardiac structures in non-contrast CT (NCCT) could improve cardiovascular risk stratification. However, setting a manual reference to train a fully convolutional network (FCN) for automatic segmentation of NCCT images is hardly feasible, and an FCN trained on coronary CT angiography (CCTA) images would not generalize to NCCT. Therefore, we propose to train an FCN with virtual non-contrast (VNC) images from a dual-layer detector CT scanner and a reference standard obtained on perfectly aligned CCTA images.