Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Seeing Beyond the Image: ECG and Anatomical Knowledge-Guided Myocardial Scar Segmentation from Late Gadolinium-Enhanced Images

Farheen Ramzan, Yusuf Kiberu, Nikesh Jathanna, Meryem Jabrane, Vicente Grau, Shahnaz Jamil-Copley, Richard H. Clayton, Chen, Chen

TL;DR

This work tackles automatic myocardial scar segmentation from LGE-MRI, a task challenged by variable contrast and artifacts. It introduces a multimodal framework that fuses ECG-derived electrophysiology with an AHA-17 anatomical prior through a Temporal Aware Feature Fusion (TAFF) mechanism to align non-simultaneous ECG and MRI data. On 103 paired MRI–ECG cases, the method achieves a Dice score of $0.8463$ (versus $0.6149$ for the image-only baseline) with precision $0.9115$ and sensitivity $0.9043$, demonstrating substantial gains from incorporating physiological and anatomical cues. Ablation studies show that both the anatomical prior and time-conditioned fusion are essential for performance, underscoring the value of physiologically grounded cross-modal segmentation and suggesting avenues for extending to 3D processing and digital ECGs.

Abstract

Accurate segmentation of myocardial scar from late gadolinium enhanced (LGE) cardiac MRI is essential for evaluating tissue viability, yet remains challenging due to variable contrast and imaging artifacts. Electrocardiogram (ECG) signals provide complementary physiological information, as conduction abnormalities can help localize or suggest scarred myocardial regions. In this work, we propose a novel multimodal framework that integrates ECG-derived electrophysiological information with anatomical priors from the AHA-17 atlas for physiologically consistent LGE-based scar segmentation. As ECGs and LGE-MRIs are not acquired simultaneously, we introduce a Temporal Aware Feature Fusion (TAFF) mechanism that dynamically weights and fuses features based on their acquisition time difference. Our method was evaluated on a clinical dataset and achieved substantial gains over the state-of-the-art image-only baseline (nnU-Net), increasing the average Dice score for scars from 0.6149 to 0.8463 and achieving high performance in both precision (0.9115) and sensitivity (0.9043). These results show that integrating physiological and anatomical knowledge allows the model to "see beyond the image", setting a new direction for robust and physiologically grounded cardiac scar segmentation.

Seeing Beyond the Image: ECG and Anatomical Knowledge-Guided Myocardial Scar Segmentation from Late Gadolinium-Enhanced Images

TL;DR

This work tackles automatic myocardial scar segmentation from LGE-MRI, a task challenged by variable contrast and artifacts. It introduces a multimodal framework that fuses ECG-derived electrophysiology with an AHA-17 anatomical prior through a Temporal Aware Feature Fusion (TAFF) mechanism to align non-simultaneous ECG and MRI data. On 103 paired MRI–ECG cases, the method achieves a Dice score of (versus for the image-only baseline) with precision and sensitivity , demonstrating substantial gains from incorporating physiological and anatomical cues. Ablation studies show that both the anatomical prior and time-conditioned fusion are essential for performance, underscoring the value of physiologically grounded cross-modal segmentation and suggesting avenues for extending to 3D processing and digital ECGs.

Abstract

Accurate segmentation of myocardial scar from late gadolinium enhanced (LGE) cardiac MRI is essential for evaluating tissue viability, yet remains challenging due to variable contrast and imaging artifacts. Electrocardiogram (ECG) signals provide complementary physiological information, as conduction abnormalities can help localize or suggest scarred myocardial regions. In this work, we propose a novel multimodal framework that integrates ECG-derived electrophysiological information with anatomical priors from the AHA-17 atlas for physiologically consistent LGE-based scar segmentation. As ECGs and LGE-MRIs are not acquired simultaneously, we introduce a Temporal Aware Feature Fusion (TAFF) mechanism that dynamically weights and fuses features based on their acquisition time difference. Our method was evaluated on a clinical dataset and achieved substantial gains over the state-of-the-art image-only baseline (nnU-Net), increasing the average Dice score for scars from 0.6149 to 0.8463 and achieving high performance in both precision (0.9115) and sensitivity (0.9043). These results show that integrating physiological and anatomical knowledge allows the model to "see beyond the image", setting a new direction for robust and physiologically grounded cardiac scar segmentation.

Paper Structure

This paper contains 11 sections, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. Multi-modal Network Architecture
  4. TAFF: Temporal Aware Feature Fusion
  5. AHA-17 Segment as Anatomical Prior
  6. Loss Function
  7. Experiments
  8. Results and Analysis
  9. Conclusion
  10. Compliance with ethical standards
  11. Acknowledgments

Figures (3)

  • Figure 1: Overview of the proposed (a) multi-modal network combining LGE-MRI with AHA-based spatial prior, and ECG features via separate image and ECG autoencoders. (b) Temporal-Aware Feature Fusion (TAFF), where a gating network adaptively integrates ECG and MRI features, modulated by the ECG–MRI acquisition interval.
  • Figure 2: Result Visualization: ground truth (a) and segmentation predictions of left ventricular blood pool, healthy myocardium, and scar overlaid on a 2D LGE slice using unimodal (image-only) (b) and the proposed multimodal approach (c). We also plot the scar volume distribution on the AHA-17 segments accordingly (d-f); and cross-lead (g) and temporal lead-wise (h) attention maps for the corresponding ECG input using the ECG attention network to support model explainability.
  • Figure 3: Mean Dice across time-interval bins for models w/ and w/o time conditioning.