ReEvalMed: Rethinking Medical Report Evaluation by Aligning Metrics with Real-World Clinical Judgment
This work addresses the gap between automated evaluation and clinician trust in medical report generation, which is crucial for improving reliability in healthcare applications.
The paper tackled the problem that existing metrics for evaluating automatically generated radiology reports do not align with real-world clinical judgment, and proposed a clinically grounded Meta-Evaluation framework to systematically assess and reveal limitations in current metrics, such as failing to distinguish clinically significant errors.
Automatically generated radiology reports often receive high scores from existing evaluation metrics but fail to earn clinicians' trust. This gap reveals fundamental flaws in how current metrics assess the quality of generated reports. We rethink the design and evaluation of these metrics and propose a clinically grounded Meta-Evaluation framework. We define clinically grounded criteria spanning clinical alignment and key metric capabilities, including discrimination, robustness, and monotonicity. Using a fine-grained dataset of ground truth and rewritten report pairs annotated with error types, clinical significance labels, and explanations, we systematically evaluate existing metrics and reveal their limitations in interpreting clinical semantics, such as failing to distinguish clinically significant errors, over-penalizing harmless variations, and lacking consistency across error severity levels. Our framework offers guidance for building more clinically reliable evaluation methods.