Abstracts

Rationale: Dense array EEG (dEEG, ≥128 channels) is increasingly used in both research and clinical applications for improved spatial resolution and higher yield of epileptiform abnormalities compared to standard 19-27 electrode EEG. Since most software does not incorporate all 256 channels in spike and seizure detection algorithms, visual review remains the gold standard for detection of epileptiform activity. Since only 30-40 channels can be viewed simultaneously, visual review of all 256 channels can be very inefficient.Methods: The 256 channel EEG net (Geodesic Sensor Net, Electrical Geodesics, Inc.) includes electrode locations approximating the International 10-20 and 10-10 systems, but intervening electrodes do not have a standardized location, and many electrodes over the cheeks and face do not have positions defined by the 10-10 or 10-5 systems. While electrode locations can be determined by digitizing the 3 dimensional position of each electrode, this is not useful for visual review. We therefore divided the geodesic net into 16 anatomical regions indicating their location on the scalp or face (Figure 1). The proposed montage incorporates standard 10-10 labels as reference points, prefixes representing regions of head and face, and the associated net electrode number to ensure accurate co-registration of sensors for source localization.Results: Our initial review strategy was to perform visual analysis and automated spike detection in a subsampled average reference montage including only named 10-20 or 10-10 electrode positions. When suspected interictal eptilepiform activity was seen, the data was expanded to include the entire 256 channel array. Since this 256 channel data was not organized anatomically and only 30-40 channels could be viewed simultaneously, it was difficult to determine the field of epileptiform activity and other normal EEG transients such as vertex waves, lambda waves, and positive occipital sharp transients of sleep. In addition, many epileptiform discharges from the inferior temporal and frontal regions were completely missed in 10-10 montages. Our newly defined montage allowed rapid identification and localization of epileptiform activity in full 256 channel views. Figure 2 demonstrates interictal data collected using a 256 channel Geodesic Sensor Net, recorded with a common vertex reference and digitally referenced to various montages for review. Figure 2 (a) interictal spike shown in standard bipolar montage, “double banana”. Figure 2 (b) is the same spike data shown in subset of electrodes from our proposed dEEG montage that clearly display the large field of the spike not visible in the standard montage.Conclusions: A standard viewing montage for dEEG both boosts efficiency when reviewing data, and also improves accuracy in recognizing epileptiform discharges. Additionally, this proposed standard dEEG montage could help direct future development of spike detection programs to recognize the discharge propagation patterns within these anatomical cluster groups, at the dense array level.