Neural Angular Plaque Characterization: Automated Quantification of Polar Distribution for Plaque Composition
Abstract: Automated quantification of plaque attenuation and calcification angle within in-vivo coronary arteries using intravascular ultrasound (IVUS) is essential for risk stratification. However, due to the physical limitations of ultrasound, the pixel-level plaque characterization has difficulties in practical application. To overcome these barriers, we propose a novel approach called Neural Plaque Angular Characterizer (NeuPAC), which learns to map an IVUS frame to the polar distribution of plaque composition. This approach inherits the advantages of both labeling and modeling. NeuPAC uses a rough label representing polar information rather than an accurate plaque composition map. NeuPAC directly outputs angular information, which is essential to clinical decision-making. Our empirical results show that NeuPAC performs well in recognizing high-risk coronary lesions for assisting clinicians in real-time.
0 Replies
Loading