Medial temporal SEEG pairs had longer aperiodic-knee timescales than lateral pairs in 206 RAM patients

In 206 patients with epilepsy from the RAM study, bipolar stereo-electroencephalography (SEEG) pairs classified as medial temporal by MNI-coordinate boxes had longer fitted aperiodic-knee timescales than lateral temporal pairs from the same patient, matched over the same y and z ranges. For the median log10 timescale difference (medial minus lateral), the mean was +0.41 (95% CI +0.35 to +0.47); 83% of patients had a positive difference (two-sided sign-flip p = 5e-5; Holm-adjusted p = 1e-4). Median knee frequencies were 2.8 Hz medial and 7.2 Hz lateral. The medial aperiodic exponent was lower by 1.03 (95% CI -1.14 to -0.91; 7% of patients positive; Holm-adjusted p = 1e-4), indicating flatter medial spectra. The result was similar for z = -25 to -10 (D_tau +0.43, n = 193, 84% positive, p = 5e-5). In a post hoc analysis excluding fits at a grid edge, D_tau was +0.28 (n = 193, 85% positive, p = 5e-5). The classes are coordinate boxes rather than atlas or imaging labels; the medial group includes hippocampus, amygdala, parahippocampal tissue, and white matter. Fits used 2–40 Hz, and many medial knees were near or below the 2 Hz boundary. Knee and exponent were fitted together and trade off, so the knee difference may partly reflect spectral shape. Each patient contributed one 60-second task segment; all had epilepsy, there was no seizure-onset-zone exclusion, and bipolar inter-contact distance was not controlled. The findings are consistent with the prediction that limbic cortex has longer timescales than association cortex, within these measurement and sampling limits.

stereo-electroencephalography · aperiodic knee · timescales · temporal lobe · RAM study · epilepsy · spectral analysis