RICAU-Net: Residual-block Inspired Coordinate Attention U-Net for Segmentation of Small and Sparse Calcium Lesions in Cardiac CT

2024-09-11

arXiv.org

AI Summary

Accurate segmentation of tiny, sparse coronary calcifications in cardiac CT remains challenging, leading to unreliable vessel-specific Agatston scoring. To address this, we propose an enhanced U-Net architecture incorporating a residual-inspired dual-path coordinate attention module—featuring dual positional embeddings and feature recalibration—and a customized Combo Loss (Dice + Focal) to mitigate severe class imbalance. Our method achieves high-precision lesion-level localization and markedly improves robustness for small-object detection. Evaluated on four-vessel coronary calcification segmentation, it attains the highest lesion-level Dice score among six compared models—including five state-of-the-art medical U-Net variants—demonstrating consistent superiority across all vessels. This work provides a reliable, automated framework for vessel-specific coronary calcium quantification, advancing clinical decision support in cardiovascular risk assessment.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Segmentation
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

The Agatston score, which is the sum of the calcification in the four main coronary arteries, has been widely used in the diagnosis of coronary artery disease (CAD). However, many studies have emphasized the importance of the vessel-specific Agatston score, as calcification in a specific vessel is significantly correlated with the occurrence of coronary heart disease (CHD). In this paper, we propose the Residual-block Inspired Coordinate Attention U-Net (RICAU-Net), which incorporates coordinate attention in two distinct manners and a customized combo loss function for lesion-specific coronary artery calcium (CAC) segmentation. This approach aims to tackle the high class-imbalance issue associated with small and sparse CAC lesions. Experimental results and the ablation study demonstrate that the proposed method outperforms the five other U-Net based methods used in medical applications, by achieving the highest per-lesion Dice scores across all four lesions.

Problem

• Cardiac Calcification Detection
• Computerized Tomography Imaging
• Coronary Artery Disease Diagnosis

Innovation

• RICAU-Net
• residual blocks
• coordinate attention