Tunnel morph model for time frequency bio-signal waveform processing

In this paper, we proposed a tunnel morph model for bio-signal waveform in measuring their similarity. Firstly, the formal specifications of bio-signal waveforms are given. And then, a series of model establishing related definitions are presented. These definitions contain waveform segmentation; waveforms distance measurement, and tunnel width computation. Moreover, on the base of the model, a similarity measuring strategy which takes the curve feature of bio-signal into account was presented. In the end, the strategy was compared with other similarity measurement methods by AECG (Ambulatory Electrocardiogram) waveform data. The data are adopted from MIT/BIH arrhythmia database. Experiment results show that the sensitivity and the positive predictivity of the strategy based on tunnel morph model are prior to other strategies.