Fast detection of venous air embolism in Doppler heart sound using the wavelet transform

Abstract

The introduction of air bubbles into the systemic circulation can result in significant morbidity. Real-time monitoring of continuous heart sound in patients detected by precordial Doppler ultrasound is, thus, vital for early detection of venous air embolism (VAE) during surgery. In this study, the multiscale feature of wavelet transforms (WT's) is exploited to examine the embolic Doppler heart sound (DHS) during intravenous air injections in dogs. As both humans and dogs share similar physiological conditions, the authors' methods and results for dogs are expected to be applicable to humans. The WT of DHS at scale 2 j(j=1,2) selectively magnified the power of embolic, but not the normal, heart sound. Statistically, the enhanced embolic power was found to be sensitive (P<0.01 at 0.01 ml of injected air) and correlated significantly (P<0.0005, τ=0.83) with the volume of injected air from 0.01 to 0.10 ml. A fast detection algorithm of O(N) complexity with unit complexity constant for VAE was developed (processing speed=8 ms per heartbeat), which confirmed the feasibility of real-time processing for both humans and dogs.