Robust and Explainable Bicuspid Aortic Valve Diagnosis Using Stacked Ensembles on Echocardiography

2026-05-13

AI Summary

This study addresses the performance instability of transthoracic echocardiography (TTE) in diagnosing bicuspid aortic valve (BAV), which arises from variability in operator expertise and image quality. To overcome this challenge, the authors propose a multi-backbone video-stacked ensemble deep learning model that leverages routinely acquired parasternal long-axis (PLAX) cine clips to enable high-accuracy automatic classification between BAV and tricuspid aortic valve (TAV). The method innovatively integrates frame-level Grad-CAM with global SHAP values to provide case-level interpretability, ensuring transparent and auditable model decisions. Evaluated using leakage-aware stratified outer-fold cross-validation, the model achieves an F1 score of 0.907 and a recall of 0.877 on external testing, demonstrating robust reliability and generalizability in real-world clinical settings.

Abstract

Transthoracic echocardiography (TTE) is the first-line imaging modality for diagnosing bicuspid aortic valve (BAV), yet diagnostic performance varies with operator expertise and image quality. We developed an explainable AI model that distinguishes BAV from tricuspid aortic valves (TAV) using routinely acquired parasternal long-axis (PLAX) cine loops. A multi-backbone video ensemble was trained and evaluated using a leakage-aware, stratified outer cross-validation protocol on $N{=}90$ patient studies (48 BAV, 42 TAV). Across fixed outer splits and 10 random seeds, the calibrated stacked ensemble achieved an outer-CV F1-score of $0.907$ and recall of $0.877$. Frame-level Grad-CAM localized salient evidence to the aortic root and leaflet plane, while globally aggregated SHAP values quantified each video backbone's contribution to the stacked prediction, enabling transparent, case-level auditability. These findings indicate that PLAX-based video ensembles can support reliable BAV/TAV classification from routine echocardiographic cine loops and may facilitate earlier detection in non-specialist or resource-limited clinical settings.

Problem

• bicuspid aortic valve
• transthoracic echocardiography
• diagnostic variability
• PLAX cine loops
• BAV/TAV classification

Innovation

• stacked ensemble
• explainable AI
• echocardiography video analysis
• Grad-CAM
• SHAP