FPGAs powering AI in medical devices

FPGA technology has transformed the landscape of medical devices and diagnostics through its exceptional capabilities. By leveraging real-time processing, high performance computing, and customisable hardware architecture, FPGAs have revolutionised medical technology, enabling more precise diagnostics and enhanced patient care outcomes.

Due to the parallel processing capabilities of FPGAs, many tasks can be carried out concurrently.

Applications using medical imaging that require the speedy processing of large amounts of data can benefit from this capacity. The efficiency of medical picture reconstruction and analysis is improved by parallel processing, which also cuts down on overall processing time. FPGAs enable AI-driven analysis and allow for processing data directly at the Edge, providing personalised treatment recommendations based on real-time insights. This on-device processing reduces latency and also enhances privacy.

In the AI field, FPGAs are often used as AI accelerators and AI processors that help enable AI workloads from Edge to Cloud. The interconnectivity within an FPGA resembles the neural wiring in the human brain. The programmable logic fabric within an FPGA is similarly connected, which is one reason why FPGAs are excellent implementation targets for neural networks and other AI workloads.

Key applications of FPGAs in medical devices:

• Medical imaging: ultra-sound/CT scanners/MRI machines
• Patient monitoring: ECG and EEG monitors
• Diagnostics and treatment: remote surgeries

Case study: ultra-low latency streamer

A ULL Streamer has been built by iWave around the Zynq UltraScale+ MPSoC FPGA. The ULL Streamer minimises this glass-to-glass delay to just 66ms over Wi-Fi 5, enabling real-time visualisation and analysis at both local and remote locations.

The FPGA powering the ULL streamer features HDMI RX and TX and Wi-Fi 5, enabling seamless data reception and transmission for real-time viewing, processing, and integration with displays. The Zynq UltraScale+ MPSoC is particularly efficient for video streaming purposes due to its integrated H.264/H.265 video codec unit. This feature allows for hardware-accelerated video encoding and decoding, making it highly suitable for high-definition video streaming.

The Encoder captures high-quality video from 12G-SDI or HDMI 2.0 sources at up to 3840 x 2160p resolution and 60 frames per second. The video is captured during surgical procedures, encoded into smaller data packets and then streamed over Wi-Fi 5 through protocols like UDP, SRP, RTMP, and RTSP. The encoded data is sent to a decoder, which displays the video on high-speed displays like Apple Vision Pro.

The ULL Streamer can be used for applications such as telemedicine, remote surgery, medical training, monitoring and simulations.

Advantages of an FPGA system on module in product development

An FPGA system on module contains all the essential components such as the FPGA, high-speed DDR memory, flash storage, power management, and interface controllers. A SoM supports high-speed transceiver blocks and multiple communication protocols, including Ethernet, USB, and PCIe, ensuring seamless connectivity and integration into various systems.

A SoM based approach involves a shorter development time, and fewer mistakes during the development cycle which can help get to market faster. The system on modules are all pre-tested and qualified by manufacturers such as iWave and enables designers to implement faster in their products. Companies can focus on their unique value proposition and core competencies, rather than worry about the design complexity of the system on modules.

The SoM approach for the FPGA SoCs further allows greater scalability for the end applications in terms of logic density, FPGA IOs, and a number of transceiver lanes. For example, a well-designed carrier board design architecture can cover system IO ports for multiple end products ranging from the FPGA with 192K logic cells to an FPGA with 1.1M logic cells. Also, the SoM approach allows migrating new generation SoC solutions without changing the product mechanical architecture and key carrier board design. This allows for lower development effort compared to a full hardware redesign from scratch.

An FPGA system design involves hundreds of components, which requires the supply chain and procurement teams to interface with a large number of suppliers and co-ordinate pricing and delivery. Through a SoM approach, iWave and other SoM companies take the responsibility of maintaining a steady supply chain through forecasting with key suppliers and maintain a steady supply. This helps in better production lead times and better product pricing, since the SoM vendors are better able to manage economies of scale.

The SoMs are all available with a pre validated BSP and design examples that can help you get started quickly. With system on modules, you can develop application software on the same production-ready hardware that you develop on, without needing to port your design from a development kit.

This article originally appeared in the March'25 magazine issue of Electronic Specifier Design – see ES's Magazine Archives for more featured publications.