A Method for Characteristic Extraction of Ultrasound Doppler Signal with Peak-valley Relationship under Heavy Noise

Abstract

This paper proposes a real-time algorithm estimating clinical useful parameters from the maximum frequency curve of ultrasound Doppler spectrum. Traditional methods always apply some pre-calculations to the frequency waveform, for instance, filtering or scaling transformation, which have limitations on real-time features detection and waveform display. In this paper, we propose and maintain a process unit as a section of the waveform with the help of its phase information and the idea of the changeable scaling in the wavelet, to determine the Doppler waveform parameters in real time. From a set of in vivo Doppler waveforms, our proposed algorithm can pick up right parameters even in very noisy environment.