The effects of protocol factors and participant characteristics on functional near-infrared spectroscopy data quality after stroke

Abstract

Functional Near-Infrared Spectroscopy (fNIRS) is an emerging neurotechnology that has several advantages over fMRI, but questions remain about factors that affect data quality and activity in stroke survivors. We examined the effect of protocol factors (Aim 1) and participant characteristics (Aim 2) on raw fNIRS signal quality and tested associations between quality control metrics and brain activity and connectivity (Aim 3) in a sample of 107 individuals with a history of left or right hemisphere stroke. Participants completed tasks that varied by cognitive and motor speech demands (from low to high): Resting State, Discourse Comprehension, and Picture Naming. Scalp-coupling indices, peak spectral power values, and number of bad channels from each task were extracted from the Quality Testing of Near Infrared Scans (QT-NIRS) toolbox (Montero-Hernandez and Pollonini, 2020) and used to index raw data quality. Data quality did not vary by session location or protocol experience, but all data quality metrics from Picture Naming were significantly lower than those from the other tasks. fNIRS signals were generally worse for Black women compared to Black men and White individuals regardless of gender. No significant associations between the raw fNIRS signal quality and Resting State functional connectivity were found. However, relative changes in Picture Naming hemoglobin concentrations were associated with scalp-coupling indices for certain channels. These results highlight the need for careful data preprocessing of already collected data and a systematic approach in future studies to mitigate inherent biases of optical instruments, thereby enhancing the inclusion of underrepresented groups in neuroscience research.