Automated analysis of transmitral and aortic Doppler velocity profiles using audio signal processing

Spectral Doppler echocardiography has been accepted as the standard non-invasive diagnostic procedure for evaluating cardiac hemodynamics. However, clinical ability to draw conclusions from derived parameters is dependent on manual tracings of the Doppler flow patterns. These manual tracings are cumbersome and subjective, with very poor reproducibility. A new automated signal analysis and envelope detection method has been developed which can trace continuous as well as pulsed Doppler flows and offers reliable results while eliminating inter-observer variability. Audio Doppler signals are acquired from normal subjects simultaneous with routine video storage of the spectral velocity data. A signal processing technique highlighted by a method of flow generation using short-time Fourier transforms provides a digital Doppler flow pattern. Customized software, using the modal velocity as a guide, has been developed for envelope detection and parameter estimation of such digital signals. These results have been compared with the observer traced and estimated parameters.