A high frequency oscillation rhythm detection method for optimizing a Gaussian mixture model on the basis of quantization errors

The invention provides a high frequency oscillation rhythm detection method for optimizing a Gaussian mixture model on the basis of quantization errors. The method includes the steps of detecting highfrequency oscillation rhythms on the basis of a clustering analysis method, selecting fuzzy entropy, short-term energy, power ratio and spectral centroid as features of epilepsy electroencephalogramsignals, forming feature vectors by the features and taking the feature vectors as an input of a clustering algorithm, classifying the feature vectors through an expected maximization Gaussian mixturemodel clustering algorithm, establishing a quantization error model, optimizing the number of clusters, improving the detection precision of the high frequency oscillation rhythms of the epilepsy electroencephalogram signals, selecting a median and an interquartile range to analyze the statistical features of each class, and detecting the high frequency oscillation rhythm. A high frequency oscillation rhythm detecting devi