Automatyczna analiza aktywności elektrycznej mózgu podczas snu i czuwania wskazuje źródło napadów w lekoopornej epilepsji
Automated intracranial spike analysis across sleep-wake states localizes the seizure onset zone in drug-resistant epilepsy
W skrócie
Badacze analizowali zarejestrowaną aktywność elektryczną mózgu pacjentów z epilepsją, którą zmienili leki. Odkryli, że specjalna metoda automatycznego wyszukiwania nieprawidłowych wyładowań elektrycznych może dokładnie wskazać miejsce w mózgu, gdzie zaczynają się napady, szczególnie podczas fazy drugiej snu. Ta metoda może pomóc lekarzom w planowaniu operacji, ponieważ działa zarówno u dorosłych, jak i u dzieci, i nie wymaga analitycznego rozpatrywania różnych faz snu.
Oryginalny abstract (angielski)
OBJECTIVE: Interictal epileptiform discharges (IEDs) are an established biomarker of epilepsy, but their utility for seizure-onset zone (SOZ) localization has been limited by reliance on short recording samples, research-grade detectors, and a primary focus on spike frequency. We assessed multi-day intracranial IED activity across sleep-wake states and its ability to localize the SOZ in adults and children. METHODS: We analyzed simultaneous scalp and intracranial EEG data from 17 patients in the EPILEPSIAE database (predominantly adult) and 10 pediatric patients at the Alberta Children's Hospital (ACH), with 48 to 72 h of data per patient. FDA-cleared clinical software (Persyst P15a) performed automated sleep staging and IED detection. We introduced a novel IED activity metric reflecting IED amplitude and the relative consistency of channel participation across detected events and used logistic regression with leave-one-patient-out cross-validation to predict SOZ channels for each sleep-wake state. RESULTS: IED activity was significantly higher in SOZ than in non-SOZ channels across every sleep-wake state in both cohorts. N2 sleep yielded the highest AUROC for SOZ localization (0.69 EPILEPSIAE; 0.77 ACH), but differences across states were not statistically significant. The presence of clinical seizures during the recording did not significantly affect localization performance. CONCLUSIONS: Automated intracranial IED analysis localizes the SOZ in adults and children, with only modest variation across sleep-wake states. SIGNIFICANCE: Amplitude-aware IED analysis using clinically deployed software is a practical adjunct to presurgical SOZ localization. The modest effect of sleep-wake states suggests that clinical value can be obtained without state-specific analysis. This approach may strengthen presurgical evaluation in adult and pediatric drug-resistant epilepsy.