Wrist Pulse Feature Variability Analysis via Spectral Decomposition

Wrist pulse waveform analysis is common in traditional Chinese medicine (TCM) engineering and diagnosis modernization. Time-domain features extracted from wrist pulse wavefrom trains oscillate with respect to time. An application of spectral decomposition, i.e., spectral independent component analysis (ICA), to the variability analysis of these feature series is proposed in this paper. Time-relevant and magnitude-relevant feature series collections of controls and patients are analyzed by spectral ICA to extract dominant spectral components, which give the indications of the VLF (very low frequency), LF (low frequency), HF (high frequency) characteristics and relevant power contributions. The results from analyzing short-term real data show the feasibility of the proposed spectral decomposition method and indicate the potential of further applications to syndrome classification and diagnosis.