Noninvasive, automated and reliable detection of spreading depolarizations in severe traumatic brain injury using scalp EEG

Abstract

AbstractBackgroundNoninvasive detection of spreading depolarizations (SD), as a potentially treatable mechanism of worsening brain injuries after traumatic brain injuries (TBI), has remained elusive. Current methods to detect SDs are based on intracranial recording, an invasive method with limited spatial coverage. Less invasive methods to diagnose SD are needed to improve generalizability and application of this emerging science and to guide worsening brain injury treatments. Here, we demonstrate, for the first time, a signal processing paradigm that can enable automated detection of SDs using noninvasive electroencephalography (EEG).MethodsBuilding on our previously developed WAVEFRONT algorithm, we have designed a novel automated SD detection method. This algorithm, with learnable parameters and improved velocity estimation, extracts and tracks propagating power depressions, as well as near-DC shifts using low-density EEG. This modified WAVEFRONT is robust to the amplitude outliers and non-propagating depressions on the scalp. We show the feasibility of detecting SD events (700 total SDs) in continuous, low-density scalp EEG recording (95±42.2 hours with 19 electrodes) acquired from 12 severe TBI patients who underwent decompressive hemicraniectomy (DHC) and intracranial EEG that could be used as a ground truth for event detection. We quantify the performance of WAVEFRONT in terms of SD detection accuracy, including true positive rate (TPR) and false positive rate (FPR), as well as the accuracy of estimating the frequency of SDs.ResultsWAVEFRONT achieves the best average validation accuracy of 74% TPR (with 95% confidence interval of 70.8%-76.7%), with less than 1.5% FPR using Delta band EEG. Preliminary evidence suggests that WAVEFRONT can achieve a very good performance (regression with R2 ≃0.71) in the estimation of SD frequencies.ConclusionsWe demonstrate feasibility and quantify the performance of noninvasive SD detection after severe TBI using an automated algorithm. WAVEFRONT can potentially be used for diagnosis and monitoring of worsening brain injuries to guide treatments by providing a measure of SD frequency. Extension of these results to patients with intact skulls requires further study.