Authors :
Presenting Author: hu lingli, MD – Epilepsy Center, Department of Neurology, Second Affiliated Hospital, School of Medicine, Zhejiang University
liu xiaochen, no – Department of Information and Electronic Engineering, Zhejiang University; ye hongyi, no – Epilepsy Center, Department of Neurology, Second Affiliated Hospital, School of Medicine, Zhejiang University,; ye lingqi, no – Epilepsy Center, Department of Neurology, Second Affiliated Hospital, School of Medicine, Zhejiang University,; wang shuang, no – Epilepsy Center, Department of Neurology, Second Affiliated Hospital, School of Medicine, Zhejiang University
Rationale:
The identification of the epileptogenic zone (EZ), defined as the region of the beginning and primary organization of ictal discharges, is essential for planning epilepsy surgery in drug-resistant focal epilepsy.
Current quantitative methods for measuring the EZ boundary usually depend on fast activity at early seizure onset. Here we hypothesize that the harmonic pattern (H pattern), a spectral structure property of ictal intracranial EEG, can help EZ localization.
Methods:
We consecutively included 70 patients with drug-resistant focal epilepsy who underwent stereo-EEG evaluation and resective epilepsy surgery. All the patients showed focal onset pattern by visual identification. Time-frequency maps were generated using Morlet wavelet transform analysis and further validated by multitaper method. H pattern was defined as multiple equidistant high-density frequency bands in the time-frequency map. Seizure-onset zone (SOZ), propagation zone (PZ) and non-involved zone (NIZ) and seizure onset pattern (SOP) were defined by experienced neurophysiologist. The parameters of H pattern were extracted in
Brainstorm. A threshold of Q3 was employed to confirm channels expressing dominant H pattern. Bicoherence analysis and mathematical simulation were used to analyze the generative mechanism underlying the multiple equidistant frequency bands.
Results:
Of the 70 patients with various SOP, H pattern was observed in 57 (81.4%) patients. H pattern appeared 18.1 ± 20.5 s after seizure onset. H pattern was consistently associated with periodic spike (PS-H pattern) or fast activity (FA-H pattern).
The start and end time of FA-H pattern were both earlier than PS-H pattern. The minimum and maximum frequency, number of bands and common difference of FA-H pattern were higher than those of PS-H pattern. H pattern occurred in SOZ, PZ and NIZ, but the distribution proportion and numbers of bands of H pattern was highest in SOZ, and the distribution proportion and maximum frequency were lowest in NIZ.
No differences were found in the start and end time, or minimum frequency of H pattern
across the three zones.
Patients with seizure freedom had a significantly larger resected ratio of areas with dominant H pattern compared with those with seizure recurrence. No such difference was observed for the SOZ. On multivariate analysis, the
resected ratio of areas with
dominant H pattern was significantly associated with prognosis. The results of the bicoherence analysis and mathematical simulations verified that the equidistant frequency bands shown in H pattern could be successively generated order by order via nonlinear wave-wave couplings.
Conclusions:
H pattern universally presents on time-frequency maps of ictal EEG in focal epilepsy, reflecting nonlinear wave-wave interaction in epileptic tissues instead of “artifact.” Presence of dominant H pattern, independent of various SOP and seizure evolution stage, helps to localize the EZ. Spectral structure of ictal EEG provides additional localizing information and new insight into the epileptogenic network.
Funding:
This work was supported by the National Natural Science Foundation of China (grant numbers: 81971207, 82171437, 81971208, 82001365).