FineRadScore: A Radiology Report Line-by-Line Evaluation Technique Generating Corrections with Severity Scores

TL;DR

FineRadScore is an automated, line-by-line evaluation tool for chest X-ray reports that uses a large language model to assess correction severity and generate explanations, aligning well with radiologist judgments.

Contribution

This work introduces FineRadScore, a novel LLM-based metric that quantifies report differences with severity scores and explanations, improving automated evaluation of CXR reports.

Findings

FineRadScore's corrections match radiologist opinions.

It aligns with radiologists and state-of-the-art metrics in report quality assessment.

Provides detailed, explainable evaluation of generated reports.

Abstract

The current gold standard for evaluating generated chest x-ray (CXR) reports is through radiologist annotations. However, this process can be extremely time-consuming and costly, especially when evaluating large numbers of reports. In this work, we present FineRadScore, a Large Language Model (LLM)-based automated evaluation metric for generated CXR reports. Given a candidate report and a ground-truth report, FineRadScore gives the minimum number of line-by-line corrections required to go from the candidate to the ground-truth report. Additionally, FineRadScore provides an error severity rating with each correction and generates comments explaining why the correction was needed. We demonstrate that FineRadScore's corrections and error severity scores align with radiologist opinions. We also show that, when used to judge the quality of the report as a whole, FineRadScore aligns with radiologists as well as current state-of-the-art automated CXR evaluation metrics. Finally, we analyze FineRadScore's shortcomings to provide suggestions for future improvements.