Leveraging Fine-Tuned Large Language Models for Interpretable Pancreatic Cystic Lesion Feature Extraction and Risk Categorization

2025-07-26

AI Summary

Manual annotation of radiology reports for pancreatic cystic lesions (PCLs) is time-intensive and hinders large-scale clinical research. Method: We propose a large language model (LLM)-based framework for automated PCL information extraction and risk stratification, combining chain-of-thought prompting with lightweight fine-tuning. High-quality, reasoning-augmented training data were synthesized using GPT-4o; open-source LLMs (e.g., LLaMA, DeepSeek) were fine-tuned via QLoRA; and risk categories were mapped to clinical guidelines (e.g., Fukuoka, AGA). Contribution/Results: Evaluated on 285 MRI/CT reports, our method achieves 97–98% accuracy in extracting key morphological features and an F1-score of 0.95 for risk classification. Inter-model agreement aligns statistically with expert radiologists. To our knowledge, this is the first fully automated, interpretable, and clinically aligned PCL structuring approach that eliminates manual annotation—enhancing scalability for research and accelerating clinical translation.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Computer Vision: Large Vision Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Background: Manual extraction of pancreatic cystic lesion (PCL) features from radiology reports is labor-intensive, limiting large-scale studies needed to advance PCL research. Purpose: To develop and evaluate large language models (LLMs) that automatically extract PCL features from MRI/CT reports and assign risk categories based on guidelines. Materials and Methods: We curated a training dataset of 6,000 abdominal MRI/CT reports (2005-2024) from 5,134 patients that described PCLs. Labels were generated by GPT-4o using chain-of-thought (CoT) prompting to extract PCL and main pancreatic duct features. Two open-source LLMs were fine-tuned using QLoRA on GPT-4o-generated CoT data. Features were mapped to risk categories per institutional guideline based on the 2017 ACR White Paper. Evaluation was performed on 285 held-out human-annotated reports. Model outputs for 100 cases were independently reviewed by three radiologists. Feature extraction was evaluated using exact match accuracy, risk categorization with macro-averaged F1 score, and radiologist-model agreement with Fleiss' Kappa. Results: CoT fine-tuning improved feature extraction accuracy for LLaMA (80% to 97%) and DeepSeek (79% to 98%), matching GPT-4o (97%). Risk categorization F1 scores also improved (LLaMA: 0.95; DeepSeek: 0.94), closely matching GPT-4o (0.97), with no statistically significant differences. Radiologist inter-reader agreement was high (Fleiss' Kappa = 0.888) and showed no statistically significant difference with the addition of DeepSeek-FT-CoT (Fleiss' Kappa = 0.893) or GPT-CoT (Fleiss' Kappa = 0.897), indicating that both models achieved agreement levels on par with radiologists. Conclusion: Fine-tuned open-source LLMs with CoT supervision enable accurate, interpretable, and efficient phenotyping for large-scale PCL research, achieving performance comparable to GPT-4o.

Problem

• Automate PCL feature extraction from radiology reports
• Improve risk categorization using fine-tuned LLMs
• Achieve radiologist-level agreement in PCL analysis

Innovation

• Fine-tuned LLMs for pancreatic cyst feature extraction
• Chain-of-thought prompting enhances model interpretability
• QLoRA optimizes open-source model performance efficiently