Tissue matching did not reduce the interictal HFA SOZ AUC in 32 HUP patients

This planned analysis tested whether grey matter (GM) and white matter (WM) position explains interictal high-frequency activity (HFA, 80–150 Hz log power) separation of seizure onset zone (SOZ) bipolar SEEG pairs. It used 3,247 pairs from 32 HUP patients and labelled pair midpoints in a fsaverage aparc+aseg atlas using 3×3×3 voxel cubes. The atlas labels carried tissue information: 1–40 Hz power was higher in GM than WM pairs (per-patient AUC 0.686, 30 of 32 patients). SOZ pairs were more often in GM (GM share 40% vs 23%; AUC of GM fraction 0.630, Holm p=0.0022), and HFA was also higher in GM than WM among non-SOZ pairs (exploratory AUC 0.653). However, HFA SOZ AUC did not decrease when SOZ pairs were compared only with non-SOZ pairs in the same tissue class: 0.633 vs 0.627, difference +0.005 (95% CI −0.013 to 0.023), p=0.57. HFA separated SOZ inside WM (AUC 0.634, p=0.006). Inside GM, AUC was 0.603 (p=0.043), but this was not significant after Holm correction (p=0.085). Spike rate did not differ between GM and WM; line length had a small tissue component (+0.019). These results indicate that tissue imbalance did not explain the HFA SOZ separation in this cohort. Limitations include template coordinates with possible errors of some millimetres, one center, clinical SOZ labels, and bipolar midpoint labels. The analyzed source files are identified by recorded URLs and SHA-256 hashes; the acquisition record does not establish an exact OpenNeuro release.

epilepsy · stereo-EEG · interictal high-frequency activity · seizure onset zone · grey matter · white matter · tissue confounding · HUP · OpenNeuro ds004100 · AUC