Brain fingerprinting using EEG graph inference

Abstract

Taking advantage of the human brain functional connectome as an individual's fingerprint has attracted great research in recent years. Conventionally, Pearson correlation between regional time-courses is used as a pairwise measure for each edge weight of the connectome. Building upon recent advances in graph signal processing, we propose here to estimate the graph structure as a whole by considering all time-courses at once. Using data from two publicly available datasets, we show the superior performance of such learned brain graphs over correlation-based functional connectomes in characterizing an individual.

Joint subject-identification and task-decoding from inferred functional brain graphs via a multi-task neural network

“…The secondary objective of this work is to present a novel calssification architecture that jointly performs subject-identification and task-decoding using a single trained model. Subject-identification is conventionally performed via building an identifiability matrix via correlations between test and re-test FC graphs [8, 9, 10], whereas for task-decoding, using support vector machines [11, 5] is the more common approach. In this study, we leverage the power of neural networks and develop a multi-task neural network (MTNN) model, which optimizes for multiple losses simultaneously, offering advantages such as enhanced generalizability and reduced overfitting [12].…”