EEG binarization for burst suppression segmentation

초록

We developed an EEG burst suppression binarization method for accurate segmentation. Adaptive thresholds decided by the constant false alarm rate algorithm were employed for binarization, and the features for quantitative EEG analysis were revisited over binary data. Burst suppression segmentation were conducted by maximum likelihood estimation with Gaussian distribution models. For performance evaluation, we evaluated the accordance to the visual scores of burst suppression. The accuracies of the proposed methods were higher than the methods before binarization and the conventional burst detection methods. Moreover, the simple calculation and optimization of the proposed methods appear to be beneficial. © 2018 IEEE.