Chun Lin Li, PhD | Capital Medical University; Xu Zhang, PhD; Min Fu Yang, MD; Xian Tao Song, MD; Ying Liang, PhD; Jing Wei, PhD; Li Wei Sun, PhD; Xin Ling Geng, PhD; Xia Li, PhD; Miao He, MS; Zhong Tian Guan, PhD; Yu Bo Liu, MS; Yu Shuang Liu, MD; Bi Xi Chen, PhD
Competition Sponsor: Chinese Academy of Medical Sciences
Awardee Year: 2024
Coronary Heart Disease (CHD) is the leading cause of death worldwide, often progressing without symptoms until severe cardiovascular events occur. Early diagnosis is crucial to prevent plaque rupture, thrombosis, and associated complications such as myocardial infarction. This project focuses on CHD early diagnosis using multimodal data by developing key tissue segmentation models to extract coronary artery and epicardial fatty structures and identify novel biomarkers. Building on this, an early diagnostic model for CHD is proposed, incorporating a progressive fusion algorithm to leverage complementary information from multimodal data. The approach captures key heterogeneous features that differentiate CHD patients from healthy individuals, enabling precise diagnosis. Furthermore, a migration paradigm is designed to adapt the diagnostic model to new datasets under unlabelled conditions, addressing challenges in clinical deployment. This paradigm facilitates the practical application of deep learning models in clinical settings. Overall, the project provides an innovative method for predicting structural and functional changes in coronary arteries, contributing to the early diagnosis and management of CHD.