Proof-of-concept validation of noninvasive detection of cortical spreading depolarization with high-resolution direct current-electroencephalography with future device recommendations

Cortical spreading depolarization (SD) is increasingly recognized as a major contributor to secondary brain injury. Noninvasive SD monitoring would enable the institution of SD-based therapeutics. Our primary objective is to establish proof-of-concept validation that scalp direct-current (DC)-potentials can provide noninvasive SD detection by comparing scalp DC-shifts from a high-density electrode array to SDs detected by gold-standard electrocorticography (ECoG). Our secondary objective is to assess usability and artifact tolerance. An 83 x 58 mm thermoplastic elastomer array with 29 6-mm diameter Ag/AgCl 1-cm spaced electrodes, the CerebroPatch™ Proof-of-Concept Prototype, was adhesively placed on the forehead with an intervening electrode gel interface to record DC-electroencephalography (DC-EEG) in normal volunteers and severe acute brain injury patients in the neuro-intensive care unit some with and some without invasive ECoG electrodes. The scalp and ECoG voltages were collected by a Moberg® Advanced ICU Amplifier. Artifacts were visually identified and usability issues were recorded. SD was scored on ECoG based on DC-shifts with associated high-frequency suppression and propagation. A six-parameter Gaussian plus quadratic baseline model was used to estimate ECoG and scalp electrode time-courses and scalp-voltage heat-map movies. The similarity of the noninvasive scalp and invasive ECoG DC-shift time-courses was compared via the Gaussian fit parameters and confirmed if the Coefficient-of-Determination was >0.80. Usability and artifact issues obscured most scalp Prototype device data of the 140 ECoG-coded SDs during 11 days in one sub-arachnoid hemorrhage patient. Twenty-six of these DC-shifts were in readable, artifact-free portions of scalp recordings and 24 of these had a >0.80 Coefficient-of-Determination (0.98 [0.02], median [IQR]) between invasive ECoG and noninvasive Prototype device DC-shifts. Reconstructed heat-map movies of the scalp DC-potentials showed a 5-cm extent, -460 µV peak region that persisted for ~70 seconds. These data suggest that these scalp DC-shifts (peak -457 ± 69 µV [mean ± StD], full-width-half-maximum 70.9 ± 5.92 seconds, area 18.7 ± 2.76 cm²) depicted in the heat-map movies represent noninvasively detected SDs. These results using 26 SDs as the observational units suggest that noninvasive SD detection is possible using scalp DC-potential signals with a high spatial resolution EEG array. Although the high artifact burden data and low usability records were limiting, negative results, they serve as an important entrepreneurial recipe that provides suggestions for a future, re-designed device that would reduce artifacts and improve usability for DC-EEG SD detection needed to enable multi-modal monitoring for secondary brain injury.