See van Klink and Zijlmans (doi:10.1093/brain/awz321) for a scientific commentary on this article. Velmuruganet al. report that detecting and localizing high‐frequency oscillations (HFOs: 80–200 Hz) with MEG can improve presurgical assessment and postsurgical outcome prediction in epilepsy. Source localization of HFOs identifies an epileptogenic region with accuracy of 75%. When such localized sources are surgically resected, patients have an approximately 80% probability of achieving seizure freedom.

SummaryObjective: Specificity of ictal high-frequency oscillations (HFOs) in identifying epileptogenic abnormality is significant, compared to the spikes and interictal HFOs. The objectives of the study were to detect and to localize ictal HFOs by magnetoencephalography (MEG) for identifying the seizure onset zone (SOZ), evaluate the cortical excitability from preictal to ictal transition, and establish HFO concordance rates with other modalities and postsurgical resection. Methods: Sixty-seven patients with drug-resistant epilepsy had at least 1 spontaneous seizure each during MEG acquisition, and analysis was carried out on 20 seizures from 20 patients. Ictal MEG data were bandpass filtered (80-200 Hz) to visualize, review, and analyze the HFOs co-occurring with ictal spikes. Source montages were generated on both hemispheres, mean fast Fourier transform was computed on virtual time series for determining the preictal to ictal spectral power transition, and source reconstruction was performed with sLORETA and beamformers. The concordance rates of ictal MEG HFOs (SOZ) was estimated with 4 reference epileptogenic regions. Results: In each subject, transient bursts of high-frequency oscillatory cycles, distinct from the background activity, were observed in the periictal continuum.Time-frequency analysis showed significant spectral power surge (85-160 Hz) during ictal state (P < .05) compared to preictal state, but there was no variation in the peak HFO frequencies (P > .05) for each subgroup and at each source montage. HFO source localization was consistent between algorithms (k = 0.857 AE 0.138), with presumed epileptogenic zone (EZ) comparable to other modalities. In patients who underwent surgery (n = 6), MEG HFO SOZ was concordant with the presumed EZ and the surgical resection site (100%), and all were seizure-free during follow-up. Significance: HFOs could be detected in the MEG periictal state, and its sources were accurately localized. During preictal to ictal transition, HFOs exhibited

BACKGROUND AND PURPOSE:The frequency of seizures is an important factor that can alter functional brain connectivity. Analysis of this factor in patients with epilepsy is complex because of disease-and medication-induced confounders. Because patients with hot-water epilepsy generally are not on long-term drug therapy, we used seed-based connectivity analysis in these patients to assess connectivity changes associated with seizure frequency without confounding from antiepileptic drugs.

Objectives Experimental models have provided compelling evidence for the existence of neural networks in temporal lobe epilepsy (TLE). To identify and validate the possible existence of resting-state "epilepsy networks," we used machine learning methods on resting-state functional magnetic resonance imaging (rsfMRI) data from 42 individuals with TLE. Methods Probabilistic independent component analysis (PICA) was applied to rsfMRI data from 132 subjects (42 TLE patients + 90 healthy controls) and 88 independent components (ICs) were obtained following standard procedures. Elastic net-selected features were used as inputs to support vector machine (SVM). The strengths of the top 10 networks were correlated with clinical features to obtain "rsfMRI epilepsy networks." Results SVM could classify individuals with epilepsy with 97.5% accuracy (sensitivity = 100%, specificity = 94.4%). Ten networks with the highest ranking were found in the frontal, perisylvian, cingulo-insular, posterior-quadrant, thalamic, cerebellothalamic, and temporo-thalamic regions. The posterior-quadrant, cerebello-thalamic, thalamic, medial-visual, and perisylvian networks revealed significant correlation (r > 0.40) with age at onset of seizures, the frequency of seizures, duration of illness, and a number of anti-epileptic drugs. Conclusions IC-derived rsfMRI networks contain epilepsy-related networks and machine learning methods are useful in identifying these networks in vivo. Increased network strength with disease progression in these "rsfMRI epilepsy networks" could reflect epileptogenesis in TLE. Key Points• ICA of resting-state fMRI carries disease-specific information about epilepsy.• Machine learning can classify these components with 97.5% accuracy.• "Subject-specific epilepsy networks" could quantify "epileptogenesis" in vivo.

Microstate alterations are present in patients with TLE. This feature might be useful in the diagnosis of epilepsy even in the absence of an IED.