Editor’s Note: This content is contributed by Subh Bhattacharya, Lead Marketing of Healthcare Sciences & Medical Devices at Xilinx, Inc


Xilinx® Versal™ silicon architecture and software tools provides a way to drastically improve image quality, speed, and accuracy in medical ultrasound systems using advanced imaging techniques. This greatly improves ultrasound-based diagnostic ability in complicated procedures.

White paper WP520.png


In the medical imaging world, there is demand to significantly improve the quality of the output image from medical imaging equipment like medical ultrasound, CT scanners, etc.

Medical ultrasound, also known as diagnostic medical sonography, is an imaging method that uses high-frequency sound waves to produce images of structures within your body. The images provide valuable information for diagnosing and treating a variety of conditions. Conventional ultrasound displays the images in thin, flat sections of the body.


Medical ultrasound is the most widely accepted form of diagnostic imaging today because of many significant advantages:

  • It uses low-energy acoustic waves, and there are no known harmful side-effects on patients—unlike potential ionizing radiation from X-rays or CT scans
  • Ultrasound can capture dynamic soft issue images, which X-rays cannot
  • Ultrasound systems can be compact for desktop and portable use or can be easily transportable within a hospital environment on a cart

For years, Xilinx technology has been at the heart of signal acquisition, conditioning, processing, and beamforming in medical ultrasound. Beamforming is the shaping of the spatial distribution of the pressure field amplitude in the area of interest, and then the recombination of the received ultrasound signals to generate images for visualization and analysis. 


To address these challenges, the Xilinx architects and our partner Dr. Jorgen Jensen, renowned ultrasound scientist, looked at how to achieve better image quality, speed, and accuracy in medical ultrasound using advanced imaging techniques on Xilinx technology.

We specifically looked at techniques like ‘Synthetic Aperture (SA) Imaging’  (widely used in radar technology) and ‘Plane Wave (PA) Imaging,’ because they offer substantial frame rate improvement and accuracy for hard-to-manage diagnostic and surgical procedures like cardiac wall motion and cardiac surgery. These two methods offer a radical departure from the conventional sequential acquisition of ultrasound images, where one image line is acquired at one time. The new methods are highly parallel and can thereby reconstruct a full image in one emission. This enables super-fast imaging with thousands of frames per second, improved focusing, and penetration, providing solutions for applications like fast cardiac imaging.

Example illustration of SA imaging.png

<Figure 1 - – Example illustration of SA imaging. The first row shows the emitted wave, which is either spherical or plane. The second row depicts the beamformed low-resolution images resulting from each emission, and the bottom row shows the resulting high-resolution image from summing all the lower resolution images in phase.>

The advanced imaging techniques create several hundred times more processing demands than traditional imaging and has so far prevented any practical implementation. Implementing the dataflow for these modalities on a traditional microprocessor presents issues related to parallelism and data throughput.

Xilinx recently introduced processing chips and a new development environment for easier and real-time implementation of these advanced imaging schemes. The Xilinx Versal Adaptive Compute Acceleration Platform (ACAP) devices, as well as Xilinx’s Alveo™ data center accelerator cards, can be deployed in workstation or servers, and are the recommended systems for enabling PW and SA.

Xilinx Versal ACAP Architecture.png

<Figure 2 - Xilinx® Versal™ Adaptive Compute Acceleration Platform (ACAP) Architecture>


Xilinx Versal ACAP (Figure 2 above) is a heterogeneous architecture with a dedicated SIMD-VLIW unit (a tile processor) called an AI Engine. AI Engines contain a scalar unit, a vector unit, load units, and a memory interface, and can implement all the different structures required by the data flow algorithms in SA and PW. Versal ACAP also has a fully integrated high-speed full-blocking crossbar switch that is used for managing the extraordinary bandwidth required.

To learn how to implement advanced imaging techniques in your next-generation medical ultrasound using a single-chip implementation of the Xilinx Versal ACAP family and the associated software framework, please read the new Xilinx whitepaper, and share with your colleagues in the industry.


Medical Ultrasound Whitepaper:  Synthetic Aperture and Plane Wave Ultrasound Imaging with Versal ACAP.


Learn more about Healthcare & Industrial IoT Cyber Solutions Powered by Xilinx -




Original Date: 04-22-2020