Similarity measurement study for Bio-signal waveform based on tunnel morph

Since Bio-signal contain rich diagnosis information for medical treatment, the modeling, featuring, classification and diagnostic analysis of the similarity measurement on such waveforms bear great significance in medical research. Getting inspiration from anti-aliasing analysis in signal processing field, a tunnel morph outcome model is presented for different types of disease related to bio-signal waveform. One similarity measurement strategy based on tunnel morph model is proposed in the paper. Instead of linear value measuring on waveforms, the curve characteristics of waveforms are also taken in consideration. Along with the tunnel morph model, a set of integrity definitions are given in the paper, including segmentation, measurement function selection, tunnel morph width, and best partition point assigning. The similarity measurement strategy is validated through a case study on AECG (Ambulatory Electrocardiogram). The data are from MIT/BIH, each of them contains 30 minutes ECG waveform. In the experiment, the sensitivity and positive predictivity are of the evaluation standard on different strategies. From the result, the effectiveness of tunnel morph strategy is higher than city block distance, Euclidean distance and correlation coefficient. And the results also support the advantage of similarity measurement based on tunnel morph model.