Real-Time UltraFast Ultrasound Imaging with Xilinx Versal Architecture

A decade of technological improvement in System-On-Chip (SoC) embedded technology using Field Programmable Gate Arrays (FPGAs) have made it possible to successfully productize traditional ultrasound systems, including portable and point-of-care. However, practical implementations of advanced modalities in UltraFast Imaging like Synthetic Aperture (SA) and Plane Wave (PW) with their many advantages have not been successful because of the sheer compute performance requirements for the ultrasound pipelines in real-time. Very large vector processers (DSPs and GPUs) solves some of the problems, but it runs into traditional scaling challenges due to inflexible and inefficient memory usage and high-power consumption. Traditional FPGA solutions provide a programmable memory hierarchy, but the hardware development flow has been a barrier to broad, high-volume adoption. The need for low power and high performance is explored beyond the conventional “one size fits all” CPU scalar processing solution for Ultrafast imaging to be a viable option. This paper presents a new technology called ‘Adaptive Compute Acceleration Platform’ (ACAP) on Xilinx’s Versal™ system, which has the capability of creating a commercial Ultrasound real time system with UltraFast imaging.

In this webinar, learn about the following:

  • The realization of the SA and PW beamformer, including the interpolation and digital signal processing on Versal™ for Ultrafast B-mode and Flow imaging
  • The development process and tools using computational graph which are fixed at compile-time and associated C++ and Python classes to produce the SA and PW design
  • Versal™ can achieve a peak performance for typical Ultrasound operations of 3200 GOPs allowing the real-time implementation of UltraFast Imaging
  • Frame rates in the range of 1000 fps can be achieved with single beamformer in a 64 channel architecture on a single Versal using 50% of the available AI Engines in Versal