Development Of Multi-channel Pulse Data Acquisition System And Classification Of Seven Common TCM Pulse Conditions

Pulse diagnosis, as one of the non-invasive method of Traditional Chinese Medicine (TCM), has been a popular method of examining disease for thousands of years. But Traditional Chinese Medicine doctors usually make recipes according to their own experience, which is a block to the development of TCM. In order to improve the treatment performance, and to get the acknowledgement all over the world, it is urgent to make the TCM objective, standard, more accurate, and flexible to use in the clinical medical treatment.This paper puts forward a design proposal of pulse signal acquisition system boasting the function of wrist pulse signal taking. The sensor BP300T and PVDF combined with air pump contraled the gas wrist strap to get the best pressure of three conditions for signal taking, which analoged the TCM diagnoses. The instrument combined amplification, band-pass filter circuit, A/D conversion as well as serial communication module. The PC gets the signals and then plots it simultaneously by the system developed by the C#software. To classify the7different types of28kinds of common pulses recorded in TCM, zero-phase filter, signal segmentation and average waveform estimation preprocessing have been used and43features of time and frequency domain are extracted, then dimension reduced to5dimensions based on Isomap algorithm. Ultimately, these features will be applied to7kinds of classification methods by the proposed Two-step method.The results show that pulse signal acquisition system can regulate pulse-taking pressure automatically to get the best pressure of three conditions and plot the pulse signal as well as save the data. In the classification which combined7kinds of classification methods, Isomap algorithm shows more object and runs faster than the Variability analysis. Compared with PCA algorithm, Isomap algorithm can reduce the vector space to a lower dimension. According to the result of similarity analysis between each concurrent pulse and its relevant mono-pulses, the improved two-step classification method is proposed. Results show that the overall classification performance has been improved by10%.