KPD Based Signal Preprocessing Algorithm for Pulse Diagnosis

In the traditional Chinese medicine (TCM) wrist pulse diagnosis plays a major role in detecting the health status of an individual. It depends strongly on the doctors' long-term experience as well as on their different inferences. Being subjective and based on long-term experience, pulse detection methods are difficult to standardize. Kim pulse diagnosis (KPD), established by Wei Jin, is an efficient method validated by both traditional Chinese medicine and in recent years also by western medicine. The key step to automatic implementation of KPD, using signal processing and analysis, is the developed KPD signal acquisition device. However, the raw wrist pulse signal acquired from KPD device includes a significant amount of noise. This paper proposes several preprocessing algorithms for pulse diagnosis, that includes wavelet transform and Gaussian filter to remove noise and the iterative sliding window (ISW) algorithm to remove the baseline wander and split the continuous signal into single periods. Experimental results show, that the algorithm for baseline wander removal is efficient and that the segmented signal matches the signal described in KPD.