Federated Learning for Coronary Artery Plaque Detection in Atherosclerosis Using IVUS Imaging: A Multi-Hospital Collaboration

TL;DR

This paper tackles the challenge of automated IVUS plaque detection under privacy constraints that hinder cross-hospital data sharing. It introduces a parallel two-stage segmentation framework trained via federated learning (FedAvg) with a coordinate conversion of IVUS images from Cartesian to polar space, enabling real-time plaque delineation by computing plaque as the difference between EEM and lumen masks. The approach achieves high accuracy for EEM and lumen segmentation (~0.89–0.90 DSC) and competitive plaque segmentation (~0.70 DSC) across a three-hospital, 151-patient dataset, with 10 federated rounds and 3 clients; it also demonstrates strong agreement with expert measurements via Bland-Altman analysis. The method offers practical benefits for multi-center collaboration and privacy-preserving medical image analysis, with potential extensions to other imaging modalities and personalized risk assessment.

Abstract

The traditional interpretation of Intravascular Ultrasound (IVUS) images during Percutaneous Coronary Intervention (PCI) is time-intensive and inconsistent, relying heavily on physician expertise. Regulatory restrictions and privacy concerns further hinder data integration across hospital systems, complicating collaborative analysis. To address these challenges, a parallel 2D U-Net model with a multi-stage segmentation architecture has been developed, utilizing federated learning to enable secure data analysis across institutions while preserving privacy. The model segments plaques by identifying and subtracting the External Elastic Membrane (EEM) and lumen areas, with preprocessing converting Cartesian to polar coordinates for improved computational efficiency. Achieving a Dice Similarity Coefficient (DSC) of 0.706, the model effectively identifies plaques and detects circular boundaries in real-time. Collaborative efforts with domain experts enhance plaque burden interpretation through precise quantitative measurements. Future advancements may involve integrating advanced federated learning techniques and expanding datasets to further improve performance and applicability. This adaptable technology holds promise for environments handling sensitive, distributed data, offering potential to optimize outcomes in medical imaging and intervention.

Figures (8)

• Figure 1: Fragmentary boundaries after edge detection on IVUS images
• Figure 2: Federated learning framework among multi-hospitals.
• Figure 3: Proposed parallel plaque segmentation diagram.
• Figure 4: (A) Original 2D IVUS image (B) EEM mask (C) lumen mask in polar coordinate system.
• Figure 5: Cardiovascular appearance in the longitudinal view.