Detecting mild traumatic brain injury with MEG, normative modelling and machine learning

Diagnosis of mild traumatic brain injury (mTBI) is challenging, as the symptoms are diverse and nonspecific. Electrophysiological studies have discovered several promising indicators of mTBI that could serve as objective markers of brain injury, but we are still lacking a diagnostic tool that could translate these findings into a real clinical application. Here, we used a multivariate machine-learning approach to detect mTBI from resting-state magnetoencephalography (MEG) measurements. To address the heterogeneity of the condition, we employed a normative modeling approach and modeled MEG signal features of individual mTBI patients as deviations with respect to the normal variation. To this end, a normative dataset comprising 621 healthy participants was used to determine the variation in power spectra across the cortex. In addition, we constructed normative datasets based on age-matched subsets of the full normative data. To discriminate patients from healthy control subjects, we trained support vector machine classifiers on the quantitative deviation maps for 25 mTBI patients and 20 controls not included in the normative dataset. The best performing classifier made use of the full normative data across the entire age range. This classifier was able to distinguish patients from controls with an accuracy of 79%, which is high enough to substantially contribute to clinical decision making. Inspection of the trained model revealed that low-frequency activity in the theta frequency band (4-8 Hz) is a significant indicator of mTBI, consistent with earlier studies. The method holds promise to advance diagnosis of mTBI and identify patients for treatment and rehabilitation. Significance statementMild traumatic brain injury is extremely common, but no definite diagnostic method is yet available. Objective markers for detecting brain injury are needed to direct care to those who would best benefit from it. We present a new approach based on MEG recordings that first explicitly addresses the variability in brain dynamics within the population through normative modeling, and then applies supervised machine-learning to detect pathological deviations related to mTBI. The approach can easily be adapted to other brain disorders as well and could thus provide a basis for an automated tool for analysis of MEG/EEG towards disease-specific biomarkers.