Clinical value of magnetoencephalographic spike propagation represented by spatiotemporal source analysis: correlation with surgical outcome

Abstract

Objective: To investigate the correlation between spike propagation represented by spatiotemporal source analysis of magnetoencephalographic (MEG) spikes and surgical outcome in patients with temporal lobe epilepsy.

Methods: Thirty-seven patients were divided into mesial (n=27) and non-mesial (n=10) groups based on the presurgical evaluation. In each patient, ten ipsilateral spikes were averaged, and spatiotemporal source maps of the averaged spike were obtained by using minimum norm estimate. Regions of interest (ROIs) were created including temporoparietal, inferior frontal, mesial temporal, anterior and posterior part of the lateral temporal cortex. We extracted activation values from the source maps and the threshold was set at half of the maximum activation at the peak latency. The leading and propagated areas of the spike were defined as those ROIs with activation reaching the threshold at the earliest and at the peak latencies, respectively. Surgical outcome was assessed based on Engel's classification. Binary variables were created from leading areas (restricted to the anterior and mesial temporal ROIs or not) and from propagation areas (involving the temporoparietal ROI or not), and for surgical outcome (Class I or not). Fisher's exact test was used for significance testing.

Results: In total and mesial group, restricted anterior/mesial temporal leading areas were correlated with Class I (p<0.05). Temporoparietal propagation was correlated with Class II-IV (p<0.05). For the non-mesial group, no significant relation was found.

Conclusions: Spike propagation patterns represented by spatiotemporal source analysis of MEG spikes may provide useful information for prognostic implication in presurgical evaluation of epilepsy.

Keywords: Epilepsy surgery; Magnetoencephalography (MEG); Minimum norm estimate (MNE); Spike propagation; Temporal lobe epilepsy.

Figure 1. Schematic representation of analyzing spike propagation

(A) Top panel: Spatiotemporal source distribution of the averaged spike obtained by minimum norm estimate, projected on the lateral and mesial cortical surface of the right hemisphere. Second panel: Region of interests (ROIs) used for investigating spike propagation. ROIs are obtained from cortical parcellation. Third panel: Unit dipoles distributed on the cortical surface. Spatiotemporal source maps (top panel) consist of the estimated activation in each dipole. Fourth panel: Combined image of top, second and third panel. The activation values at each unit dipole are extracted. (B) The time course of the activation at each unit dipole is represented. The figure shows examples of anterior temporal, inferior frontal and temporoparietal ROIs, including AT1, IF1 and TP1 shown on (A). When the maximum peak (=0ms) activation within all ROIs appears at a unit dipole AT1 (shown as M), the threshold (Th) is determined by (1/2)xM. The spike activation reaches Th at the latency t(AT1). The latency “t” is obtained at each unit dipole of all ROIs in the same manner, such as t(IF1). Note that Th is a constant applied for all unit dipoles. In the figure, t(AT1) is earlier than t(IF1). ROIs containing the earliest “t” are considered leading areas of the spike. The peak activation at IF1 is larger than Th and ROIs including such unit dipoles are considered propagated areas. In contrast, the activation at TP1 does not reach Th.

Figure 2

Spatiotemporal source maps of the averaged spike in early and peak (=0ms) latency obtained from representative patients of mesial group. (A) 50-year-old female (Patient 6). Early activation appears at -14ms in the anterior temporal area. Peak activation is seen in small portions of the mesial and posterior temporal areas as well as the anterior temporal area. The patients has been seizure free after a right temporal lobectomy (Engel's Class I). (B) 18-year-old male (Patient 21). Early inferior frontal activation appears at -12ms as well as the anterior/mesial temporal activation. Peak activation involves posterior temporal and temporoparietal areas in addition to the early activation areas. The patient underwent a right temporal lobectomy but still has occasional disabling seizures (Class II).

Figure 3

Spatiotemporal source maps of the averaged spike in early and peak (=0ms) latency obtained from representative patients of non-mesial group. (A) 32-year-old male (Patient 30). Early activation at -28ms is seen in the right mid-temporal area. Peak activation extends to the whole temporal region and temporoparietal area. The patient underwent a right temporal lobectomy and only has simple partial seizures with a transient strange sensation in the head (Class I). (B) 22-year-old female (Patient 32). Early activation at -8ms involves right mid-temporal and temporoparietal areas. Peak activation is seen in the same areas with larger strength. The patient underwent a subpial topectomy with transection but still has frequent seizures (Class IV).