Research Of Hemocyte Signal Identification Base On HHT-HMM

Hemocyte analysis is an important method to diagnose disease on clinical. Hemocyte signal is nonlinear, non-stationary signal. This paper proposes an algorithm composed with Hilbert-Huang transform and hidden Markov model to identify the hemocyte between the healthy people and the hyperthyroid patients.Hilbert-Hang transform can get a high resolution in both time domain and frequency domain. The self-adaption Hilbert-Huang transform is appropriate for hemocyte analysis. This paper solve the pseudo intrinsic mode functions of empirical mode decomposition by using wavelet thresholding denoise technique. Hemocyte signal can be decomposed into a series of intrinsic mode functions by empirical mode decomposition. The IMF with high correlation coefficient is considered the true IMF. The energy moment, spectral centroid and marginal spectrum entropy is calculated as the eigenvector. A special hidden Markov model named hidden-semi Markov model is used for identification. An initial parameters closer to global optimal is calculated by an algorithm similar to k-means. The result show that it is a good way to identify the hemocyte signal between healthy people and the hyperthyroid patients.