Data Acquisition

MICA-PNI dataset

MICA-PNI dataset images were all acquired at the McConnell Brain Imaging Centre of the Montreal Neurological Institute on a 7 T Terra Siemens with a 32-receive and 8-transmit channel head coil in parallel transmission (pTX) mode. Quantitative T1 images were acquired using a three-dimensional magnetization-prepared two rapid gradient echo (MP2RAGE) sequence with universal pulses (UPs) to improve B1+ homogeneity. Imaging parameters were: 0.5 mm isotropic resolution; matrix size, 320 × 512 × 520; repetition time (TR), 5170 ms; echo time (TE), 2.81 ms; inversion times (T1​/T2​), 1000/3200 ms; flip angles, 4°; in-plane GRAPPA acceleration factor, 3; partial Fourier factor, 0.75; field of view (FOV), 160 × 256 × 260 mm3. These data were used for cortical surface reconstruction and intracortical microstructural profiling. Diffusion-weighted images were acquired using a pulsed gradient spin-echo (PGSE) echo-planar imaging (SE-EPI) multi-shell protocol comprising 110 volumes across four diffusion weightings. These included b = 0 s/mm² (n = 10), b = 300 s/mm² (n = 7), b = 1000 s/mm² (n = 28), and b = 2000 s/mm² (n = 64), and a b = 0 image acquired with reversed phase-encoding to enable correction of susceptibility-induced distortions. Imaging parameters were: 1.1-mm isotropic resolution; matrix size, 192 × 192; repetition time (TR), 4840 ms; echo time (TE), 79.4 ms; flip angle, 90°; refocusing flip angle, 180°; field of view (FOV); 211 × 211 mm²; partial Fourier factor, 0.75; multiband acceleration factor, 3; GRAPPA acceleration factor, 3. These data were used to estimate whole-brain structural connectivity. Resting-state functional MRI data were acquired using a multiecho T2*-weighted gradient-recalled echo-planar imaging (GRE-EPI) sequence (Center for Magnetic Resonance Research, University of Minnesota). Imaging parameters were: 1.9 mm isotropic resolution; matrix size, 118 × 118; 75 slices oriented 39° relative to the AC–PC line; repetition time (TR), 1690 ms; echo time (TE1/TE2/TE3), 10.8/27.3/43.8 ms; flip angle, 67°; FOV, 224 × 224 mm2; partial Fourier factor, 0.75; multiband acceleration factor, 3; GRAPPA acceleration factor, 3. During the resting-state acquisition, participants were instructed to fixate on a gray cross and remain at rest for 7 min 46 s, yielding 275 volumes.

BigBrain dataset

An ultrahigh-resolution 3D reconstruction of a sliced and cell-body-stained postmortem human brain from a 65-year-old woman was obtained from the open-access BigBrain repository (https://bigbrain.loris.ca/main.php). The postmortem brain was paraffin-embedded, coronally sliced into 7,400 20-μm sections, stained using the Merker method for cell bodies and digitized. Manual inspection for artefacts (such as rips, tears, shears and stain crystallization) was followed by automatic repair procedures, involving nonlinear alignment to a postmortem MRI of the same individual acquired before sectioning, together with intensity normalization and block averaging. The 3D reconstruction was implemented with a successive coarse-to-fine hierarchical procedure. We downloaded the 3D volume at 100-μm resolution, which was the highest resolution available for the whole brain. Computations were performed on inverted images, where intensity reflects greater cellular density and soma size.

AHEAD dataset

The Ahead brain open-access dataset is a multimodal 3D reconstruction of the whole human brain at 200 μm resolution, created by integrating ultra-high-field MRI and light microscopy from two post-mortem human brains Microscopic images were obtained coronally by sectioning the brain at 200 μm intervals and applying six different staining techniques. The ANTs SyN algorithm was used to re-register blockface and microscopic images, minimizing nonlinear distortions and ensuring accurate alignment of brain boundaries and vascular maps. Although small intensity variations were present in the 3D reconstruction, these were linearly adjusted to preserve the original dynamic range of each stain. In our study, we utilized a single subject from the dataset, a 59-year-old female (https://doi.org/10.21942/uva.16844500.v1), and selected only Bielschowsky and Parvalbumin stainings, which target nerve fibers and interneurons, respectively.

MNI open iEEG atlas

Atlas of the human intracranial electroencephalogram (Discovery cohort). The MNI open iEEG atlas is a multicentre initiative providing openly available human intracranial recordings acquired during different states of vigilance. This dataset includes iEEG recordings from a sample of 106 patients (52 women; mean standard deviation (SD) age = 33.1 ± 10.8 years) with intractable epilepsy, totalling 1772 data points taken from non-epileptic channels mapped to a common stereotactic space. The present study focused on segments recorded during resting wakefulness with eyes closed. All data was collected during a standard clinical protocol implemented in all participating sites.

MICA iEEG dataset

Intracranial EEG data from the MICA laboratory were obtained from 31 patients (sub-PX* series) with intractable epilepsy who underwent presurgical evaluation. Recordings were acquired during resting wakefulness (eyes closed, stage-W) and organized in BIDS format. Electrode contact sensitivity maps were derived from subject-specific leadfield models computed as part of the electroMICA derivatives pipeline. For each contact, sensitivity maps were rectified (absolute value), thresholded at 0.001, and aggregated across hemispheres onto the fsLR-32k surface (32,492 vertices per hemisphere) to produce a per-contact spatial weighting vector used for surface-based mapping.