A Wireless Frequency Domain Beamforming Based Ultrasound Imaging

Background: Considerable reduction both in sampling rate and processing time can be achieved in ultrasound imaging by applying compressed sensing, Xampling and frequency domain beamforming, leading to sub-Nyquist sampling rates. This study evaluated this model to create a wireless ultrasound.

Methods: Ultrasonic signals were received by a commercial ultrasound system transducer and were amplified, sampled and transformed to the frequency domain by a hardware component referred to as a "Xampler". The low rate data output received from this component was transmitted over a 802.11n wireless link to a computer where frequency domain beamforming was implemented in C language, utilizing Intel`s Integrated Performance Primitives (IPP) library to enhance performance. The design was implemented partly in Verilog and partly in VHDL. Simulations and synthesis were performed with VCS-MX and Design Compiler of Synopsys respectively. The physical design was done using Cadence`s Encounter tool.

Results: Imaging results are shown in Figure 1. An image reconstructed at the remote site from 122 complex valued samples per scan line, transmitted over a wireless link, had a quality similar to an image obtained by a state of the art commercial ultrasound imaging system with 2048 real-valued samples per scan line. The saving of 244/2048 ≈ 1/8 in raw data rate has enabled the wireless operation.

Discussion: Xampling and frequency domain beamforming manage to reduce sampling considerably enabling significant data storage and transmission reduction. Further studies are needed to correlate this system with clinical data and pathologies.