Inverse Problem for M/EEG Source Localization: A Review

This systematic review provides a comprehensive evaluation of current and emerging methodologies in magnetoencephalography/electroencephalography (M/EEG) source localization by addressing critical technological and methodological gaps. The review follows systematic review and meta-analysis (PRISMA) guidelines and meticulously examines a wide range of source localization methods, making it the most exhaustive evaluation in the field. It categorizes 28 distinct methods into five core groups: basic, hybrid, subspace-based, probabilistic, and machine learning-based, and offers an in-depth comparative analysis. Our findings reveal a pressing need for advancements in accuracy, spatial-temporal resolution, and computational efficiency. We identify the imperative for improved signal processing, advanced modeling, integration of machine learning, and AI, to enhance source localization accuracy. The review advocates for personalized, real-time applications in clinical settings and underscores the importance of multimodal neuroimaging studies for comprehensive brain activity insights. We recommend longitudinal, large-scale studies and open science practices for validating and generalizing findings. This review stands as a definitive guide for future research, aiming to propel M/EEG source localization to new heights of accuracy and clinical utility.