A pipeline for large-scale assessments of dementia EEG connectivity across multicentric settings

Multicentric initiatives based on high-density electroencephalography (hd-EEG) are urgently needed for classification and characterization of diseases subtypes in diverse and low-resource settings. These initiatives are, however, not without challenges, with sources of variability arising from differing data acquisition and harmonization methods, multiple preprocessing pipelines, and different theoretical modes and methods to compute source space/scalp functional connectivity. Our team developed a novel pipeline aimed at the harmonization of hd-EEG datasets and dementia classification. This pipeline handles data from recording to machine learning classification based on multi-metric measures of source space connectivity. A user interface is provided for those with limited background in MATLAB. Here, we present our pipeline and detail a comprehensive step-by-step example for prospective analysts reviewing the five main stages of the pipeline: data pre-processing; normalization; source transformation; connectivity metrics and dementia classification. This detailed step-by-step pipeline may improve the assessment of heterogenous, multicentric, and multi-methods approaches to functional connectivity in aging and dementia.