Ran Score: a LLM-based Evaluation Score for Radiology Report Generation

1. 1Develop a clinician-guided framework for multi-label finding extraction from free-text chest X-ray reports
2. 2Combine human expertise with large language models to improve recognition of low-prevalence abnormalities and handling of clinically important language
3. 3Define a finding-level metric, Ran Score, for report evaluation
4. 4Use Ran Score to evaluate and refine radiology report generation models

Radiologists and AI engineers on a team benefit from Ran Score as it improves the accuracy of radiology report generation and evaluation, enabling more effective collaboration between clinicians and AI systems

💡 Combining human expertise with large language models can improve the accuracy and effectiveness of radiology report generation and evaluation

📊 Introducing Ran Score: a LLM-based evaluation score for radiology report generation #AIinRadiology #LLMs