Neurophysiological Correlates of Concussion: Deep Learning for Clinical Assessment

Abstract

Concussion has been shown to leave the afflicted with significant cognitive and neurobehavioural deficits. The persistence of these deficits and their link to neurophysiological indices of cognition, as measured by event-related potentials (ERP) using electroencephalography (EEG), remains restricted to population level analyses that limit their utility in the clinical setting. In the present paper, a convolutional neural network is extended to capitalize on characteristics specific to EEG/ERP data in order to assess for post-concussive effects. An aggregated measure of single-trial performance was able to classify accurately (85%) between 26 acutely to post-acutely concussed participants and 28 healthy controls in a stratified 10-fold cross-validation design. Additionally, the model was evaluated in a longitudinal subsample of the concussed group to indicate a dissociation between the progression of EEG/ERP and that of self-reported inventories. Concordant with a number of previous studies, symptomatology was found to be uncorrelated to EEG/ERP results as assessed with the proposed models. Our results form a first-step towards the clinical integration of neurophysiological results in concussion management and motivate a multi-site validation study for a concussion assessment tool in acute and post-acute cases.

Figure 1

(A) The general study timeline and datapoints collected from each group. (B) The TRODNet architecture and sizes provided an input batch of size 1 that contains data from 64 channels and 3 experimental conditions across 332 samples. (C) The training/testing procedure accounting for both concussed subsets.

Figure 2

The interaction effect of Recovery and Testing Date on the TRODNet results as seen on the longitudinal subgroup. While there were main effects of both factors, no reliable interaction was found. Points represent mean prediction from TRODNet’s result, where 0 (1) is a classification of control (concussed). Vertical extended lines indicate the 95% confidence intervals.

Figure 3

Interactions between days since injury and symptomatology (first row), depressive symptoms (second row), and TRODNet single-trial results in the longitudinal sample of our presented dataset (third row). The symptom resolution (SR) subgroup conveyed no identifiable patterns both in the first (left column) and second (right column) tests. The subgroup that did not have symptoms resolve (NSR) showed an increase in symptomatology and depressive signs as days since injury increased for the second test. Shaded regions signify the 95% confidence intervals.

Figure 4

The mean of the absolute SHAP values for single-subject averages overlaid on the head for each condition and electrode. The abscissa denote time where 0 is the stimulus onset. The ordinate represents the mean absolute SHAP value at the indicated electrode, time, and condition. The figure shows a robust identification of ERPs of interest, particularly in the frequency (FDev) and Duration (DDev) deviants. An interesting effect can be observed to the standard condition where the parieto-occipital region has a widespread effect predominantly in the right hemisphere.