arXiv:2512.03339v1 Announce Type: cross 
Abstract: Ejection fraction (EF) is a crucial metric for assessing cardiac function and diagnosing conditions such as heart failure. Traditionally, EF estimation requires manual tracing and domain expertise, making the process time-consuming and subject to interobserver variability. Most current deep learning methods for EF prediction are black-box models with limited transparency, which reduces clinical trust. Some post-hoc explainability methods have been proposed to interpret the decision-making process after the prediction is made. However, these explanations do not guide the model's internal reasoning and therefore offer limited reliability in clinical applications. To address this, we introduce ProtoEFNet, a novel video-based prototype learning model for continuous EF regression. The model learns dynamic spatiotemporal prototypes that capture clinically meaningful cardiac motion patterns. Additionally, the proposed Prototype Angular Separation (PAS) loss enforces discriminative representations across the continuous EF spectrum. Our experiments on the EchonetDynamic dataset show that ProtoEFNet can achieve accuracy on par with its non-interpretable counterpart while providing clinically relevant insight. The ablation study shows that the proposed loss boosts performance with a 2% increase in F1 score from 77.67$\pm$2.68 to 79.64$\pm$2.10. Our source code is available at: https://github.com/DeepRCL/ProtoEF

تم تقديم نموذج جديد يسمى ProtoEFNet لتقدير كسر الإخراج (EF) في تخطيط صدى القلب، حيث يعالج هذا النموذج القيود التي تفرضها الطرق التقليدية التي تتطلب تتبعًا يدويًا وتكون عرضة للتباين. يستخدم هذا النموذج نماذج ديناميكية مكانية زمنية لالتقاط أنماط الحركة القلبية المهمة، مما يعزز دقة وتفسير تقديرات EF.

Se ha introducido un nuevo modelo llamado ProtoEFNet para estimar la fracción de eyección (EF) en ecocardiografía, abordando las limitaciones de los métodos tradicionales que requieren trazado manual y son propensos a la variabilidad. Este modelo utiliza prototipos espaciotemporales dinámicos para capturar patrones de movimiento cardíaco significativos, mejorando la precisión y la interpretabilidad de las predicciones de EF.

Un nouveau modèle nommé ProtoEFNet a été introduit pour estimer la fraction d'éjection (EF) en échocardiographie, répondant aux limites des méthodes traditionnelles qui nécessitent un traçage manuel et sont sujettes à des variations. Ce modèle utilise des prototypes spatiotemporels dynamiques pour capturer des motifs de mouvement cardiaque significatifs, améliorant ainsi la précision et l'interprétabilité des prédictions d'EF.

A new model named ProtoEFNet has been introduced for estimating ejection fraction (EF) in echocardiography, addressing the limitations of traditional methods that require manual tracing and are prone to variability. This model utilizes dynamic spatiotemporal prototypes to capture significant cardiac motion patterns, enhancing the accuracy and interpretability of EF predictions.

ProtoEFNet: Dynamic Prototype Learning for Inherently Interpretable Ejection Fraction Estimation in Echocardiography

Was this article worth reading? Share it

finlight.me

LucidQuery AI

MicroEstimates

Eyeye

Deptho.ai

E2B Dev

Ready to build your own newsroom?