Researchers from Simon Fraser University, the Hospital for Sick Children and Defense Research and Development Canada teamed together using m agnetoencephalography (MEG) imaging to develop a new technique allowing high-resolution brain scans to be paired with machine learning algorithms, which in turn improves the diagnosis of mild traumatic brain injury (mTBI).
The team's findings, published in PLOS Computational Biology, suggest that the results could be a game changer in detecting concussions. Currently, doctors identify concussions largely in part by patients reporting symptoms. With this new technique, researchers used MEG to map brain activity on participants with and without concussions.
The imaging technique, which directly measures brain activity at fast time scales, showed the changes in communication between certain regions of the brain, related to traumatic brain injury.
“We found characteristic alterations of inter-regional interactions associated with concussion,” wrote lead authors Vasily Vakorin and Sam Doesburg. “Moreover, using a machine learning approach, we were able to detect concussion with 88 percent accuracy from MEG connectivity, and confidence of classification correlated with symptom severity.”
The authors of “Detecting Mild Traumatic Brain Injury Using Resting State Magnetoencephalographic Connectivity” add that MEG imaging of resting state functional connectivity may offer new methods for detecting and assessing the severity of concussion using neuroimaging.
"An objective, quantitative method for diagnosing brain dysfunction after mTBI would allow identification of patients at risk for a subsequent injury, be invaluable for developing parameters around return to play / work / duty, and assist in developing guidelines for providing care, monitoring treatment efficacy and tracking recovery."