8 Analysis of EEG in Medically Intractable Epilepsy
181
than those of structural MRI (76% sensitivity, 53% specificity), PET (69% sensitivity,
44% specificity) and ictal SPECT (58% sensitivity, 47% specificity). The accuracy
of ESI was higher in case of temporal lobe epilepsy than extratemporal lobe epilepsy,
although the difference did not reach statistical significance. The sensitivity of ESI
in extratemporal lobe epilepsy was 75%, which was sharply contrasted to clinical
localization of scalp EEG in which the diagnostic sensitivity of IEDs was reported
to 37.1% in a large series of surgical cases of neocortical epilepsy [24].
The same group has extended their study to determine which procedure or combination of procedures is the most predictive of seizure-free outcome in patients
undergoing epilepsy surgery in 190 patients with partial epilepsy patients [23]. The
predictive value of non-invasive techniques including high-density electric source
imaging (HD-ESI), MRI, PET, and SPECT in terms of sensitivity, specificity, predictive value and diagnostic odds ratio (OR) in relation to postoperative outcome
which was evaluated at least 12 months after surgery (mean 26.6 ± 27 months).
EEGs were obtained with 64, 128, or 256 electrodes in 85 patients. Averaged spikes
were used and ESI was estimated by LAURA. Among 58 patients who underwent
all tests, the adjusted prognostic OR was 13.1 for HD-ESI and 10.9 for MRI (both
p 0.004). 92.3% were seizure-free when combination of MRI and HD-ESI were
positive, and none was seizure-free if both tests were negative.
Feng et al. [13] reported accuracy of ESI in forty-three patients with temporal lobe epilepsy (age ranged from 9 to 48 years old) [13]. ESI was estimated
from EEG recording acquired with high density 256-channel. Forward model was
a realistic atlas head model based on the finite difference method (FDM), and Low
resolution electromagnetic tomography (LORETA) method was used for inverse
estimation. Postoperative surgical outcome was used as a gold standard for epileptogenic zone (follow-up duration: minimum 7 months, median 14 months, average
13.6 months). 35 patients had good surgical outcome (Engel class I or II), and eight
patients had Engel class III or IV. Sensitivity and specificity were measured with
reference to surgical resection margin. At the sublobule level, the sensitivity and
specificity of high density ESI were 91.4% and 75%, respectively, which showed
the best performance compared to other noninvasive neuroimaging modalities (PET:
80% sensitivity and 50% specificity, MRI: 77.1% sensitivity and more than 60%
specificity). The accuracy of high density ESI was comparable to that of the previous
study [8].
It is remarkable that they investigated the relationship between the high density
ESI source patterns and surgical outcome with the Kaplan–Meier survival analysis
which showed that cases with single source had better prognosis than cases with
multiple sources (88.9% vs. 42.9%, respectively). Furthermore, surgical prognosis
was better when sources are present within the resection margin than when sources
are beyond the resection margin (94.1% vs. 33.3%, respectively). However, they did
not report how many patients had lesion on MRI, and how many patients had mesial
or lateral temporal lobe epilepsy. It was not also reported how many patients were
of children.
ESI can also be successfully applied on children with epilepsy. Russo et al. [36]
retrospectively evaluated the diagnostic utility of ESI in 60 pediatric epilepsy patients
181
than those of structural MRI (76% sensitivity, 53% specificity), PET (69% sensitivity,
44% specificity) and ictal SPECT (58% sensitivity, 47% specificity). The accuracy
of ESI was higher in case of temporal lobe epilepsy than extratemporal lobe epilepsy,
although the difference did not reach statistical significance. The sensitivity of ESI
in extratemporal lobe epilepsy was 75%, which was sharply contrasted to clinical
localization of scalp EEG in which the diagnostic sensitivity of IEDs was reported
to 37.1% in a large series of surgical cases of neocortical epilepsy [24].
The same group has extended their study to determine which procedure or combination of procedures is the most predictive of seizure-free outcome in patients
undergoing epilepsy surgery in 190 patients with partial epilepsy patients [23]. The
predictive value of non-invasive techniques including high-density electric source
imaging (HD-ESI), MRI, PET, and SPECT in terms of sensitivity, specificity, predictive value and diagnostic odds ratio (OR) in relation to postoperative outcome
which was evaluated at least 12 months after surgery (mean 26.6 ± 27 months).
EEGs were obtained with 64, 128, or 256 electrodes in 85 patients. Averaged spikes
were used and ESI was estimated by LAURA. Among 58 patients who underwent
all tests, the adjusted prognostic OR was 13.1 for HD-ESI and 10.9 for MRI (both
p 0.004). 92.3% were seizure-free when combination of MRI and HD-ESI were
positive, and none was seizure-free if both tests were negative.
Feng et al. [13] reported accuracy of ESI in forty-three patients with temporal lobe epilepsy (age ranged from 9 to 48 years old) [13]. ESI was estimated
from EEG recording acquired with high density 256-channel. Forward model was
a realistic atlas head model based on the finite difference method (FDM), and Low
resolution electromagnetic tomography (LORETA) method was used for inverse
estimation. Postoperative surgical outcome was used as a gold standard for epileptogenic zone (follow-up duration: minimum 7 months, median 14 months, average
13.6 months). 35 patients had good surgical outcome (Engel class I or II), and eight
patients had Engel class III or IV. Sensitivity and specificity were measured with
reference to surgical resection margin. At the sublobule level, the sensitivity and
specificity of high density ESI were 91.4% and 75%, respectively, which showed
the best performance compared to other noninvasive neuroimaging modalities (PET:
80% sensitivity and 50% specificity, MRI: 77.1% sensitivity and more than 60%
specificity). The accuracy of high density ESI was comparable to that of the previous
study [8].
It is remarkable that they investigated the relationship between the high density
ESI source patterns and surgical outcome with the Kaplan–Meier survival analysis
which showed that cases with single source had better prognosis than cases with
multiple sources (88.9% vs. 42.9%, respectively). Furthermore, surgical prognosis
was better when sources are present within the resection margin than when sources
are beyond the resection margin (94.1% vs. 33.3%, respectively). However, they did
not report how many patients had lesion on MRI, and how many patients had mesial
or lateral temporal lobe epilepsy. It was not also reported how many patients were
of children.
ESI can also be successfully applied on children with epilepsy. Russo et al. [36]
retrospectively evaluated the diagnostic utility of ESI in 60 pediatric epilepsy patients