Removal of wall-motion artefacts in Doppler ultrasound signals using linear prediction filtering

The authors present a new approach to filtering wall-motion artefacts from pulsed Doppler ultrasound signals. The technique estimates the vessel wall-motion signal and linearly removes it from the backscattered signal, preserving low-frequency blood flow information which would have otherwise been removed had conventional highpass filtering methods been applied. This provides a much improved representation of the blood flow profile throughout the cardiac cycle. The filter employs linear prediction to model the wall-motion signal and linearly subtracts it from the original Doppler signal, thus retaining only blood flow information.