A Unified DINOv2-Based Framework for LVEF Estimation, GLS Dysfunction Classification, and Early Cardiotoxicity Prediction

2026-08-13

AI Summary

This study addresses the challenge of cardiac function assessment without cardiac cycle segmentation in cardio-oncology by proposing a unified framework based on DINOv2. By integrating LoRA fine-tuning, temporal aggregation, and physiology-guided hybrid regression, the method innovatively enables cycle-free inference and multi-task joint learning. This approach validates the potential of visual foundation models for medical time-series analysis while significantly reducing reliance on annotated data. Experimental results demonstrate superior performance in left ventricular ejection fraction estimation (MAE 4.64%), global longitudinal strain classification (AUC 76.48%), and early cardiotoxicity prediction (AUC 70.26%). Collectively, this work establishes an efficient paradigm for precise clinical assessment in cardio-oncology, highlighting the efficacy of adapting vision foundation models to complex physiological signal interpretation without extensive manual annotation.

Abstract

Left ventricular ejection fraction (LVEF) estimation (Task 1), global longitu-dinal strain (GLS)-based dysfunction classification (Task 2), and early cardi-otoxicity prediction (Task 3) provide complementary information for cardio-oncology assessment. LVEF reflects macroscopic ventricular volume chang-es as the clinical standard, whereas GLS captures subtle myocardial defor-mation, indicating subclinical cardiotoxicity before overt LVEF decline. Fur-thermore, predicting cardiotoxicity from baseline echocardiography prior to treatment enables preventive interventions at an early stage. To address these three tasks, we employ a DINOv2-based framework with task-specific adap-tation and prediction heads. Built upon a frozen foundation encoder, the framework incorporates parameter-efficient Low-Rank Adaptation (LoRA) and temporal aggregation to learn task-specialized representations, ensuring robust generalization. Crucially, during inference, it operates in a fully cycle-detection-free and phase-free manner, requiring neither cardiac cycle seg-mentation nor explicit End-Diastolic/End-Systolic (ED/ES) annotations. Ad-ditionally, we introduce an ED/ES-guided 2D/3D hybrid multi-view regres-sion model specifically to optimize Task 1. On a patient-level split containing 1,203 training videos from 237 patients and 300 validation videos from 59 independent patients, the DINOv2-based framework achieved a mean abso-lute error (MAE) of 5.03% for Task 1, an AUC-ROC of 76.48% for Task 2, and an AUC-ROC of 70.26% for Task 3. For Task 1, the specialized ED/ES-guided model further improves performance, achieving an MAE of 4.64%. This framework demonstrates the effectiveness of foundation model repre-sentations across diverse cardio-oncology tasks and the additional benefit of physiology-guided modeling for accurate LVEF estimation.

Problem

• LVEF estimation
• GLS dysfunction classification
• early cardiotoxicity prediction
• cardio-oncology

Innovation

• DINOv2
• Low-Rank Adaptation (LoRA)
• Cycle-detection-free
• ED/ES-guided hybrid regression
• Cardio-oncology