Foundation Models Adaptation for Multi-View Multi-modal Cardiac MRI Segmentation and Direct Ejection Fraction Estimation

2026-08-07

AI Summary

This work addresses the challenge of segmenting heterogeneous multi-center, multi-sequence, and multi-view cardiac MRI images and directly estimating left ventricular ejection fraction (LVEF). The authors propose a novel approach that integrates fine-tuned and frozen foundation models for cardiac MRI. Specifically, the CineMA model is fine-tuned to achieve high-accuracy segmentation of both cine and late gadolinium enhancement (LGE) images, while frozen models extract embedding features that, combined with an attention mechanism, enable multi-instance learning for end-to-end LVEF regression. This method represents the first effective integration of multiple foundation models, overcoming the limitations of single-model approaches. Experimental results demonstrate Dice scores of 0.862–0.902 for cine and 0.621–0.846 for LGE segmentation, with an LVEF estimation mean absolute error of 4.96% and a Pearson correlation coefficient of 0.91.

Abstract

Foundation models have shown strong transferability in cardiac MRI (CMR), but their effectiveness for heterogeneous multi-view and multi-sequence CMR analysis remains unclear. In this work, we explore the effectiveness of fine-tuning and combining different CMR foundation models for the Universal Multi-Sequence, Multi-Center and Multi-View CMR Segmentation (CMR-Multi) Challenge. CineMA was fine-tuned for cine and late gadolinium enhancement (LGE) segmentation across short-axis and long-axis views. For direct left-ventricular ejection fraction (LVEF) estimation, we used two recent frozen CMR foundation models to extract embedding vectors that were then combined using attention-based multiple-instance learning for LVEF regression. In the challenge validation set, cine segmentation achieved Dice scores of 0.862, 0.883, and 0.902 for short-axis, two-chamber and four-chamber cine MRI, respectively. LGE segmentation achieved Dice scores between 0.621 and 0.846 across views. The direct LVEF regression model achieved an MAE of 4.96 percentage points and a Pearson correlation of 0.91. These results indicate that foundation models can be effectively adapted and combined for multi-view CMR analysis, while accurate LGE scar segmentation remains a challenging task.

Problem

• foundation models
• multi-view
• multi-modal
• cardiac MRI segmentation
• ejection fraction estimation

Innovation

• foundation models
• multi-view segmentation
• direct ejection fraction estimation
• attention-based multiple-instance learning
• cardiac MRI