Utility of Reduced Montage EEG in Detecting Seizures or Seizure-Like Activity
Abstract number :
1.163
Submission category :
4. Clinical Epilepsy / 4B. Clinical Diagnosis
Year :
2017
Submission ID :
334579
Source :
www.aesnet.org
Presentation date :
12/2/2017 5:02:24 PM
Published date :
Nov 20, 2017, 11:02 AM
Authors :
Kapil Gururangan, Stanford University Medical Center; Babak Razavi, Stanford University Medical Center; and Josef Parvizi, Stanford University Medical Center
Rationale: Standard full montage (>18channel) scalp electroencephalography (EEG) is used to detect a wide range of cerebral pathologies, from single epileptiform spikes to generalized seizures. However, its utility in emergency and resource-limited settings may be impeded by delays in setup and interpretation. Past studies have investigated reduced channel configurations as screening tools, but inferred a lower utility for them in detecting epileptiform abnormalities. The current study tested the utility of reduced (8-channel) montage (rm-EEG) covering the lateral hemispheres compared to full montage (fm-EEG) for detecting seizures and seizure-like patterns. Methods: Study participants were 20 neurologists with EEG training (7 epileptologists and 13 epilepsy fellows) from 7 institutions (CPMC, Kaiser Permanente, Rush, Stanford, UCLA, UCSF, Yale) and 42 medical students from Stanford University. Forty-four, 15-second long samples of EEG recordings were presented as both fm-EEG and rm-EEG formats in a random order. Samples represented seizures (focal or generalized, n=8), seizure-like activity (lateralized or generalized periodic discharges or burst suppression, n=12) or non-rhythmic, non-periodic patterns (normal or slowing, n=24) as determined by majority agreement among 3 senior epileptologists with >10 years of training (Fleiss’ kappa: 0.62 for seizure, 0.79 for seizure-like activity). Both physicians and students were asked to determine whether each sample represented seizure activity (yes/no), while epileptologists were also asked to specify any and all pathological activity present in each sample. We calculated the sensitivity and specificity of fm-EEG and rm-EEG for seizures and seizure-like activity; differences between the full montage and the reduced montage were assessed using paired t-tests. Results: Epileptologists identified seizures/seizure-like activity with 77% sensitivity and 89% specificity using rm-EEG, compared to 92% sensitivity (p < 0.001) and 83% specificity (p=0.002) using fm-EEG. Students identified seizures/seizure-like activity as seizures with 37% sensitivity and 72% specificity using rm-EEG, compared to 52% sensitivity (p < 0.001) and 62% specificity (p < 0.001) using fm-EEG. We found comparatively low sensitivity among epileptologists when they judged the presence of seizures (not seizure-like activity) on both rm-EEG (34%) and fm-EEG (41%) (p=0.08), while the specificity of rm-EEG and fm-EEG remained high at 92% and 90%, respectively (p=0.04). Students identified seizures (not seizure-like activity) with 41% sensitivity and 70% specificity using rm-EEG, compared to 58% sensitivity (p < 0.001) and 59% specificity (p < 0.001) using fm-EEG. Conclusions: Our study demonstrates that a reduction from 18 to 8 channels provides relatively high sensitivity in detecting seizures/seizure-like activity and more specific information for ruling in such epileptic activity. A restricted channel configuration can significantly reduce EEG setup time, thereby expediting diagnosis and lowering health care costs. Funding: Support provided by Stanford University Medical Scholars Research Program.
Clinical Epilepsy