Power-Efficient Retrospective Transmit Beamforming in Range-Doppler Frequency Domain

Retrospective transmit beamforming (RTB) is a synthetic aperture (SA) technique often used on clinical ultrasound (US) scanners to improve image resolution and focusing outside of the transmit focal zone. RTB involves beamforming multiple image lines per transmit and is typically implemented on a parallel processing hardware [such as graphics processing units (GPUs)] that reduces computation time but tends to increase heating and power consumption. We offer an alternative, more efficient way of performing RTB, in the range-Doppler (RD) frequency domain using the range-Doppler algorithm (RDA), which was previously shown to reduce the number of computations compared to beamforming in the time domain with delay-and-sum (DAS). Here, we adapt RDA for the first time to US signals from diverging waves and focused transmits to create RDA RTB and evaluate it using simulated, phantom, and in vivo acquisitions from a clinical US system. We show that RDA RTB can reduce CPU power consumption by 40% compared to the industry implemented and optimized DAS RTB (figure right) while achieving comparable clinical image quality and visualization of targets in abdominal scans (figure left). We also compare RDA RTB to another frequency-domain implementation of RTB based on F-k (Stolt) migration. While RDA RTB is not as power-efficient as F-k RTB (which assumes broadband signals and does not require multiple iterations to achieve optimal image quality), it consumes only half the memory required by F-k RTB since beamforming in the RD domain does not involve a Fourier transform and zero-padding in the time dimension. RDA RTB holds a significant potential to reduce power consumption on point-of-care US (POCUS) devices with limited memory and processing hardware while allowing them to achieve image quality of full-size clinical scanners.