Improving Automating Quality Control in Radiology: Leveraging Large Language Models to Extract Correlative Findings...

Area of Science:

• Medical Diagnostics
• Artificial Intelligence in Healthcare
• Radiology Reporting

Background:

• Radiology imaging is crucial for medical diagnosis, with report accuracy vital for treatment planning.
• Manual comparison of radiology and operative reports is time-consuming and requires expertise.
• Ensuring report reliability through cross-referencing is a standard but inefficient practice.

Purpose of the Study:

• To explore the use of a Large Language Model (LLM) for automating the extraction of critical information from radiology and operative reports.
• To simplify the evaluation process in radiology by leveraging AI for data extraction.
• To focus on identifying key shoulder anatomical structures within these reports.

Main Methods:

• A fine-tuned Large Language Model (LLM) was developed and applied to extract specific anatomical details.
• The LLM was trained to identify mentions of the supraspinatus tendon, infraspinatus tendon, subscapularis tendon, biceps tendon, and glenoid labrum.
• The model processed lengthy radiology and operative reports to extract relevant data.

Main Results:

• The fine-tuned LLM demonstrated a capability to accurately pinpoint relevant anatomical details within the reports.
• Initial findings suggest the model can effectively identify mentions of key shoulder structures like tendons and the labrum.
• The LLM successfully processed and extracted information from extensive medical documents.

Conclusions:

• Large Language Models show significant potential for transforming traditional radiology evaluation methods.
• Automated data extraction using LLMs can enhance the efficiency and accuracy of comparing radiology and operative findings.
• This approach offers a promising solution to the labor-intensive nature of manual report cross-referencing.