The applications of time-frequency analyses to ictal magnetoencephalography in neocortical epilepsy

Abstract

SummaryPurpose: Ictal magenetoencephalographic (MEG) discharges convey significant information about ictal onset and propagation, but there is no established method for analyzing ictal MEG. This study sought to clarify the usefulness of time-frequency analyses using short-time Fourier transform (STFT) for ictal onset and propagation of ictal MEG activity in patients with neocortical epilepsy.Methods: Four ictal MEG discharges in two patients with perirolandic epilepsy and one with frontal lobe epilepsy (FLE) were evaluated by time-frequency analyses using STFT.Prominent oscillation bands were collected manually and the magnitudes of those specific bands were superimposed on individual 3D-magnetic resonance images.Results: STFT showed specific rhythmic activities from alpha to beta bands at the magnetological onset in all four ictal MEG records. Those activities were located at the vicinity of interictal spike sources, as estimated by the single dipole method (SDM), and two of the four ictal rhythmic activities promptly propagated to ipsilateral or bilateral cerebral cortices. The patients with FLE and perirolandic epilepsy underwent frontal lobectomy and resection of primary motor area respectively including the origin of high-magnitude areas of a specific band indicated by STFT, and have been seizure free after the surgery.Conclusions: STFT for ictal MEG discharges readily demonstrated the ictal onset and propagation. These data were important for decisions on surgical procedure and extent of resection. Ictal MEG analyses using STFT could provide a powerful tool for noninvasive evaluation of ictal onset zone.

Magnetoencephalographic imaging of ictal high‐frequency oscillations (80‐200 Hz) in pharmacologically resistant focal epilepsy

“…9,35 Time-frequency analysis and source reconstruction with a spatial filter and current-density estimates are more accurate than ECD due to the advantage of analysis of a wide window rather than a single time point. However, previous studies had used either a manual selection of a particular period distinct from the BGA 36 or a wide time window. 7 The present study selected a 4-second ictal window and computed non-time-resolved frequency analysis, to circumvent the subjective bias of time window selection.…”

Resting-state fMRI studies in epilepsy

“…But one major limitation of MEG is its low sensitivity to deep brain activity. Although it has been used to detect neocortical spike discharges [5,6] , accurate localization of the epileptic focus is still a problem.…”

Usefulness of multiple frequency band source localizations in ictal MEG

“…However, several other studies reported that ictal MEG occasionally fails to localize the IOZ (Eliashiv et al, 2002;Shiraishi et al, 2001;Tilz et al, 2002). A plausible explanation suggested for this MEG failure is its low localization accuracy due to the low signal-to-noise ratio (SNR) of early spikes in ictal MEG (Fujiwara et al, 2012;Yagyu et al, 2010).…”

“…However, previous methods are limited by their subjectivity when selecting specific frequency bands and time windows for analysis. A manual selection procedure is highly dependent on the researcher opinion; thus, time window durations and frequency bands vary (Yagyu et al, 2010). Moreover, time-frequency analysis results may show no clear activity at ictal onset, which creates difficulties when selecting times and frequencies.…”

“…Some researchers have employed time-frequency analyses to determine the IOZ (Iwatani et al, 2012;Kobayashi et al, 2013;Ochi et al, 2007;Sueda et al, 2013;Yagyu et al, 2010), while others have adopted recently developed forward solution models, such as LORETA, sLORETA, MUSIC, MEG beamformer, and dSPMs, in their ictal source localization studies (Fujiwara et al, 2012;Tanaka et al, 2009). Most of the previous studies using these advanced analytical methods have reported better IOZ localization results than those obtained via conventional ECD analysis.…”