Uncertainty-Aware Structured Data Extraction from Full CMR Reports via Distilled LLMs

TL;DR

This paper introduces CMR-EXTR, a novel framework that accurately extracts structured data from free-text CMR reports and provides confidence scores for quality control, enhancing clinical data curation.

Contribution

It presents the first CMR-specific extraction system with integrated confidence estimation using a teacher-student distillation pipeline for offline inference.

Findings

Achieves 99.65% variable-level accuracy.

Effectively triages human review using uncertainty principles.

First system to combine CMR extraction with confidence scores.

Abstract

Converting free-text cardiac magnetic resonance (CMR) reports into auditable structured data remains a bottleneck for cohort assembly, longitudinal curation, and clinical decision support. We present CMR-EXTR, a lightweight framework that converts free-text CMR reports into structured data and assigns per-field confidence for quality control. A teacher-student distillation pipeline enables fully offline inference while limiting manual annotation. Uncertainty integrates three complementary principles -- distribution plausibility, sampling stability, and cross-field consistency -- to triage human review. Experiments show that CMR-EXTR achieves 99.65% variable-level accuracy, demonstrating both reliable extraction and informative confidence scores. To our knowledge, this is the first CMR-specific extraction system with integrated confidence estimation. The code is available at https://github.com/yuyi1005/CMR-EXTR.