Brain MRI Deep Learning and Bayesian Inference System Augments Radiology Resident Performance.

Area of Science:

• Medical imaging analysis
• Artificial intelligence in radiology
• Neuroradiology

Background:

• Automated quantitative and probabilistic medical image analysis offers potential improvements in radiology workflow accuracy and efficiency.
• Artificial intelligence (AI) systems combining deep learning and Bayesian networks can distinguish between 50 diagnostic entities on brain MRIs.

Purpose of the Study:

• To evaluate if AI systems can serve as a clinical decision support tool to augment radiologist performance in brain MRI diagnosis.
• To assess the impact of an AI-assisted tool on diagnostic accuracy for various levels of radiology expertise.

Main Methods:

• Utilized AI systems integrating convolutional neural networks and Bayesian networks for brain MRI analysis.
• Tested AI system performance using the Adaptive Radiology Interpretation and Education System (ARIES) in 194 cases.
• Compared diagnostic accuracy (Top Diagnosis and Top Three Differential Diagnosis) of radiologists with and without AI assistance.

Main Results:

• Radiology residents showed significant improvement with ARIES: Top Diagnosis (55% vs. 30%) and Top Three Differential Diagnosis (79% vs. 52%).
• AI assistance led to the largest improvement in diagnostic accuracy for rare diseases (39% increase in T3DDx for residents).
• Attending neuroradiologists' performance showed no significant difference with or without ARIES assistance.

Conclusions:

• A hybrid AI system (deep learning and Bayesian inference) can augment the diagnostic accuracy of non-specialists in brain MRI interpretation.
• AI-assisted decision support has the potential to elevate non-specialist diagnostic performance to near subspecialist levels for a wide range of conditions.
• The study highlights the potential of AI as a valuable tool for enhancing diagnostic capabilities in radiology, particularly for less experienced practitioners.