Wavelet Based Analysis of Doppler Ultrasonic Wrist-pulse Signals

Abstract

Traditional Chinese pulse diagnosis (TCPD) is one of the most important diagnostic techniques in Traditional Chinese Medicine (TCM) and computerized analysis of pulse signals is a crucial step in objectifying and standardizing TCPD. In this work, we use Doppler ultrasonic device to collect wrist-pulse signals from patients with gastritis and cholecystitis as well as healthy persons. After extracting the envelopes of ultrasonic pulse contour, wavelet (packet) transforms are applied to decompose the pulse signals and extract the wavelet features. Together with some Doppler ultrasonic diagnostic parameters, such as STI, RI, etc., a two-category classifier is employed to distinguish the unhealthy persons from healthy ones, and tell the patients from different diseases. 12 gastritis sufferers (Group G), 15 cholecystitis sufferers (Group C) and 19 healthy persons (Group H) were involved in the experiment. An accuracy of 80. 77% and an accuracy of 86.21% are achieved in discriminating Group G and Group C from Group H, respectively, and the classification accuracy between Group G and Group C can reach 100%.