Mutual Information Based Recognition of Pulse Signal in TCM

This study aims at utilizing mutual information algorithm to make an objective and quantitative research for the pulses in Traditional Chinese Medicine (TCM) diagnosis. The normal pulse signals, slippery pulse signals and taut pulse signals were collected from the outpatients by Shanghai University of TCM. Some typical pulse signals from all these classes were selected as the templates to be matched with the new pulse signals by means of mutual information, and then which class the new pulse signal should be classified was determined through the maximum mutual information. Finally, the satisfactory results were obtained which proof that it was appropriate to use mutual information to recognize pulse signals. The biggest advantage of this way is simple, since the inputted pulse signals should not be extracted features. And the time-domain features were extracted from the pulse signals to be classified. The results from the two different ways show the classification method presented is efficient based on Mutual Information.