Seeing Beyond the Image: ECG and Anatomical Knowledge-Guided Myocardial Scar Segmentation from Late Gadolinium-Enhanced Images
Summary: arXiv:2511.14702v4 Announce Type: replace-cross
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. A novel approach integrating electrocardiogram (ECG) signals with anatomical knowledge aims to improve this process significantly.
Introduction
Cardiac MRI is a critical tool in assessing heart conditions, particularly for identifying myocardial scarring. Traditional methods for segmenting these scars often struggle with inconsistencies in image contrast and various artifacts that can obscure the underlying tissue viability. To enhance the accuracy of myocardial scar segmentation, researchers have proposed a multimodal framework that leverages both ECG signals and anatomical information.
Methodology
The study introduces a novel framework that merges the electrophysiological data from ECGs with anatomical priors derived from the American Heart Association’s 17-segment model (AHA-17). This innovative approach is vital, as conduction abnormalities detected in ECGs can provide critical insights into the localization of scarred regions in the myocardium.
- Temporal Aware Feature Fusion (TAFF): As ECGs and LGE-MRIs are typically acquired at different times, the researchers developed a TAFF mechanism. This mechanism dynamically weights and fuses features based on the temporal differences in acquisition, allowing for a more accurate representation of the myocardial condition.
- Clinical Dataset Evaluation: The proposed method was rigorously tested on a clinical dataset, showcasing its effectiveness in comparison to existing state-of-the-art methods.
Results
The integration of ECG-derived information with anatomical knowledge resulted in substantial improvements in segmentation accuracy. The key findings include:
- Average Dice score for scar segmentation improved from 0.6149 to 0.8463.
- Precision achieved an impressive score of 0.9115.
- Sensitivity was recorded at 0.9043, highlighting robust performance.
Conclusion
The results indicate that the proposed framework allows for a more comprehensive understanding of myocardial conditions by “seeing beyond the image.” By successfully integrating physiological and anatomical knowledge, this approach sets a new direction for robust and physiologically grounded cardiac scar segmentation. Future work may explore broader applications of this methodology within the field of cardiac imaging and beyond, potentially leading to improved patient outcomes through better diagnostic accuracy.
Implications
This research not only enhances our understanding of myocardial scar segmentation but also opens pathways for future innovations in cardiac imaging technologies. By harnessing the power of multimodal data, clinicians may have access to tools that provide deeper insights into heart health, ultimately leading to more effective treatment strategies.