Researchers Use AI to Improve Brain Disorder Diagnostics

Researchers from UC San Francisco developed a machine learning algorithm to enhance 3T MRIs by synthesizing 7 Tesla (7T)-like images that approximate real 7T MRIs. The researcher’s model enhanced pathological tissue with more fidelity for clinical insights and represents a step toward evaluating clinical applications of synthetic 7T MRI models.
 
Recent studies show that ultra-high-field MRI at 7T could have greater resolution and clinical advantages over high-field MRI at 3T in delineating anatomical structures that are important for identifying and monitoring pathological tissue, particularly in the brain. 
 
UCSF researchers collected imaging data from patients diagnosed with mild traumatic brain injury (TBI) at UCSF. Three neural network models were designed and trained to perform image enhancement and 3D image segmentation using the generated synthetic-7T MRIs from the standard 3T MRIs. 
 
Senior study author Reza Abbasi-Asl, PhD, UCSF Assistant Professor of Neurology, said, 
“Our paper introduces a machine-learning model to synthesize high-quality MRIs from lower-quality images. We demonstrate how this AI system improves the visualization and identification of brain abnormalities captured by MRIs in Traumatic Brain Injury. Our findings highlight the promise of AI and machine learning to improve the quality of medical images captured by less advanced imaging systems.”
 
The images generated with the new models provided enhanced pathological tissue for patients with mild TBI. They selected an example region with white matter lesions and microbleeds in subcortical areas to use for comparison. The researchers found pathological tissue was easier to see in synthesized 7T images. 
 
The synthesized 7T images better captured the diverse features within white matter lesions. These observations also highlight the promise of using this technology to improve diagnostic accuracy in neurodegenerative disorders such as multiple sclerosis.
 
The researchers believe that future work should include extensive clinical assessment of the model findings, clinical rating of model-generated images, and quantification of uncertainties in the model.