Abstracts

Rationale: Standard 18-channel scalp electroencephalography (EEG) is used to detect a wide range of cerebral abnormalities, from subtle asymmetries and single spikes to seizures and other abnormal rhythmic periodic patterns. Past studies have investigated reduced channel configurations (e.g., 6 or 8 channels) as screening tools, but they inferred a lower utility for these restricted montages in detecting epileptic abnormalities. The current study seeks to determine the utility of reduced, 8-channel montage (RM) covering the lateral hemispheres compared to a full, 18-channel montage (FM) for detecting seizures and rhythmic periodic patterns based on brief EEG samples.Methods: We selected 44 samples, each 15-seconds long, from continuous EEG recordings acquired in standard medical care for patients with possible seizures. Half of these EEG samples represented “rhythmic periodic activity” (RPA; J Clin Neurophysiol 30(1):1, 2013). The other half represented normal activity or diffuse and focal slowing. Each EEG sample was presented in two formats – FM and RM (lateral chains of a double banana montage). After viewing training samples in FM and RM, respondents determined if subsequent samples (order randomized) represented RPA (yes/no), and if yes, its laterality (left/right/both). They also reported their confidence in their judgment (0-100% in 10% increments). The time spent on each sample was recorded. The diagnostic value of FM and RM were quantified with sensitivity, specificity, and positive and negative predictive value (PPV, NPV). A 95% confidence interval for differences in proportions (CI) was used to assess differences in FM and RM diagnostic metrics, and differences in confidence and time per sample were assessed using a two-sample t-test and a median test, respectively.Results: Three epileptologists rated the EEG samples (n=264 samples). RPA was identified in FM samples with 88% sensitivity, 80% specificity, 82% PPV and 87% NPV. By comparison, RPA was identified in RM with 77% sensitivity (CI for difference with FM [-0.02,0.23]), 91% specificity (CI [-0.22,0.01]), 98% PPV (CI [-0.20,0.04]) and 80% NPV (CI [-0.06,0.19]). For samples that were correctly judged as RPA, respondents correctly determined laterality for 84% of FM and 78% of RM samples (CI for the difference [-0.1,0.2]). Confidence in responses was 72% for FM and 70% for RM (t=0.71, p=0.48), and median time per sample was 12 sec for FM and 11 sec for RM (χ2=0.55, p=0.46).Conclusions: Scalp EEG is used routinely for detection of a wide range of abnormal brain activity including seizures and other pathological rhythmic periodic patterns. However, conventional EEG uses a large number of channels that may be more than sufficient to provide diagnostically important information. Our study demonstrates that a reduction in channel number by over 50% provides comparable diagnostic utility to full montage EEG in detecting pathological rhythmic periodic activity. A restricted channel configuration can significantly reduce EEG setup time, thereby expediting diagnosis and lowering health care costs.