Neural Fingerprinting based on Brain Network Dynamics: A Cross-Platform MEG Study

Neural fingerprinting aims to identify individuals through measurements of their distinct brain activity. Magnetoencephalography (MEG) is a potent tool for this purpose. Previous studies mostly utilized conventional SQUID-based MEG and focused on data averaged over time, omitting the intricate temporal dynamics present in MEG recordings.

In this investigation, researchers employed both SQUID-MEG and the more recent OPM-MEG techniques to demonstrate that fingerprinting was feasible both within a single modality and across different modalities using static (time-aggregated) features, findings consistent with prior research.

Furthermore, the study sought to explore whether fingerprinting could be achieved based on brain network dynamics, quantified using a canonical hidden Markov model (CHMM). The results indicated that CHMM-derived networks yielded a slightly better fit to SQUID data compared to OPM data, but this difference was minor and anticipated given that the networks were trained on SQUID data.

The researchers also discovered that fingerprinting was achievable within and between modalities using CHMM-derived state activation time courses and state power spectral densities. However, the state summary statistics and transition probabilities only facilitated within-modality fingerprinting.

The findings suggest that subject-specific information is retained within CHMM states and remains consistent across various MEG technologies. This represents a significant advancement in understanding brain dynamics, particularly in delineating brain networks as defined by the CHMM. The study also paves the way for the potential use of CHMM in processing and interpreting OPM-MEG data, opening new avenues for future research in brain network analysis.