Detection of Neonatal Amplitude-Integrated EEG Based on Revised D-S Theory

Amplitude-integrated electroencephalography (aEEG) has been widely used in continuous monitoring of neonatal brain function. This paper proposes an aEEG recognition method based on revised D-S Theory. The revised D-S Theory improves traditional D-S theory by introducing weight factor into the algorithm. Combining judgments with different weights can attenuate the conflict among them and get a more sound one. The efficiency of the proposed method is validated by classifying 103 aEEG recordings into normal and abnormal groups. Approximate entropy (ApEn) and amplitudes are used as the features to characterize aEEG signals. Compared with the traditional D-S theory, the classification accuracy of the revised method increases by 4.88%. This method could be helpful in monitoring newborn brain function.