Validating the Clinical Utility of CineECG 3D Reconstructions through Cross-Modal Feature Attribution

TL;DR

This study introduces a cross-modal mapping technique that enhances the interpretability of ECG deep learning models by projecting feature attributions onto 3D anatomical space, improving clinical relevance.

Contribution

The paper presents a novel cross-modal mapping method that aligns ECG feature attributions with 3D anatomical structures, addressing interpretability challenges in clinical ECG analysis.

Findings

Mapped explanations achieve a Dice score of 0.56, outperforming the baseline of 0.47.

Cross-modal averaging filtering improves attribution stability and localization accuracy.

Models trained directly on CineECG signals show reduced accuracy and incoherent attributions.

Abstract

Deep learning models for 12-lead electrocardiogram (ECG) analysis achieve high diagnostic performance but lack the intuitive interpretability required for clinical integration. Standard feature attribution methods are limited by the inherent difficulty in mapping abstract waveform fluctuations to physical anatomical pathologies. To resolve this, we propose a cross-modal method that projects feature attributions from high-performance 12-lead ECG models onto the CineECG 3D anatomical space. Our study reveals that while models trained directly on CineECG signals suffer from reduced accuracy and incoherent attributions, the proposed mapping mechanism effectively recovers clinically relevant feature rankings. Validated against a ground-truth dataset of 20 cases annotated by domain experts, the mapped explanations yield a Dice score of 0.56, significantly outperforming the 0.47 baseline of standard 12-lead attributions. These findings indicate that cross-modal averaging mapping effectively filters attribution instability and improves the localization of pathological features, combining the diagnostic expressiveness of standard ECG with the intuitive clarity of anatomical visualization.