A Natural Language Processing-based Model to Automate MRI Brain Protocol Selection and Prioritization.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Clinical Decision Support Systems

Background:

• Incorrect selection of imaging protocols increases healthcare costs and waste.
• Improving the quality and safety of medical imaging services is a key objective.
• Clinical decision support can optimize diagnostic processes.

Purpose of the Study:

• To develop and evaluate natural language processing (NLP) models for clinical decision support in magnetic resonance imaging (MRI) brain examinations.
• To determine the feasibility of using NLP for protocoling and prioritization of MRI brain scans.
• To assess the accuracy of NLP models in predicting examination protocols, contrast administration, and priority.

Main Methods:

• Three NLP models were developed using random forest, support vector machine, and k-nearest neighbor algorithms.
• Models predicted examination protocol, contrast need, and priority based on clinical indications and demographics.
• Data from 13,982 MRI brain examinations (Jan 2013-June 2015) were used for training and testing.

Main Results:

• Optimal model accuracy reached 82.9% for protocol selection, 83.0% for contrast administration, and 88.2% for prioritization.
• These results demonstrate the predictive capability of algorithms for aiding clinical decisions.
• The models successfully utilized narrative clinical information and demographic data.

Conclusions:

• NLP models derived from clinical notes and demographic data are feasible for predicting MRI brain examination protocols and priorities.
• These AI-driven tools can support clinical decision-making in radiology.
• The study highlights the potential of NLP to enhance efficiency and reduce errors in medical imaging.