Identifying task-activated cortex through phase-amplitude coupling analysis

Having a patient specific mapping of the eloquent cortex is crucial for patients undergoing resective surgery. Non-invasive neuroimaging modalities like magnetoencephalography (MEG) are increasingly used for pre-surgical mapping due to high temporal resolution. Phase-amplitude coupling (PAC), a form of cross-frequency coupling, involves the modulation of high-frequency amplitude by the phase of low-frequency oscillations. PAC observed during tasks reflect interactions between microscale and macroscale neural ensembles. The objective of this study was characterizing whether PAC differs during task versus spontaneous recordings to investigate the utility of PAC for mapping eloquent cortex. We analyzed MEG data from 10 subjects with somatosensory and spontaneous recordings from our retrospective database of epilepsy patients. We preprocessed the raw MEG files to eliminate noise, including cardiac and ocular artifacts. We then created epochs of 250 ms and averaged across trials. Subsequently, we constructed a head model and performed source localization using LCMV beamformer. PAC was quantified using the mean vector length modulation index on the scout-time series for 148 brain regions defined by the Destrieux atlas. Tensor decomposition using the Candecomp/Parafac algorithm facilitated data dimensionality reduction. Density-Based clustering (DBSCAN) identified anomalous regions, juxtaposed with expected task-activated regions. The same method was applied to spontaneous (resting state) data. The regions identified during the left somatosensory task directly matched the expected regions (p=1.8 x 10-9, binomial test). Conversely, no regions were identified in spontaneous data. These results demonstrate that phase-amplitude coupling is significantly altered during evoked responses and can be used as a biomarker of activated cortex.