Functional near-infrared spectroscopy pre-ictal oscillations recorded simultaneously with stereo-EEG.

Background: Peri-ictal hemodynamic alterations have been consistently demonstrated using functional near-infrared spectroscopy (fNIRS). However, whether ultra-slow hemodynamic activity develops before intracranially defined seizure onset remains uncertain. We investigated this question using simultaneous stereotactic EEG (SEEG) and fNIRS recordings, providing a precise intracranial temporal reference for seizure onset. Objective: To investigate whether ultra-slow hemodynamic activity precedes SEEG-defined seizure onset and occurs more frequently than expected by chance in prolonged recordings. Methods: Five patients undergoing simultaneous SEEG-fNIRS monitoring contributed 16 analyzable seizures. Exploratory inspection of seizure-averaged continuous wavelet scalograms identified a candidate ultra-slow frequency band(0.002-0.004 Hz), which was subsequently evaluated using a predefined permutation framework comparing true seizures with pseudo-seizure events sampled from the same recordings. Statistical inference combined channel-level permutation testing with seizure-level permutation-based Fisher analyses, supported by spatial and short-separation control analyses. Results: Evidence for ultra-slow hemodynamic activity preceding SEEG-defined seizure onset was identified in three of five patients and in 10 of 16 analyzable seizures. One long-separation channel remained significant following family-wise error correction, while multiple seizures demonstrated significant permutation-based Fisher analyses. Nominally significant channels frequently formed spatial clusters that broadly corresponded to independently defined seizure onset hypotheses and were generally not mirrored by short-separation channels. Temporal centers of gravity, used as conservative descriptors of the temporal distribution of activity, preceded SEEG-defined seizure onset by 72.0-632.1 s for HbO and 20.3-592.0 s for HbR, consistent with a substantial pre-ictal component. Conclusions: Ultra-slow hemodynamic activity frequently developed before and preferentially around SEEG-defined seizure onset compared with pseudo-seizure events, suggesting these fluctuations are unlikely to reflect chance occurrences within prolonged recordings. Although the physiological origin and clinical significance of these oscillations remain uncertain, and the analyzed frequency band should be regarded as exploratory pending independent validation, convergent channel- and seizure-level analyses, spatial organization, and limited correspondence with superficial signals support further investigation of ultra-slow hemodynamic dynamics as a component of seizure-related physiology.