Bench Talk

Source: fabioderby /stock.adobe.com; generated with AI)

Modern medical technology has an amazing impact on our health and well-being. Technologies ranging from diagnostic equipment, such as ultrasound and endoscopic cameras, to robotically assisted surgery drastically improve the level of care available today. Of course, medical electronics used for operations often require certification to the highest standards due to the potential outcome of incorrect data. These standards demand safety and security coupled with the high performance required for most end-user applications. This is where FPGAs come into play.

Many of the things that make FPGAs great for general high-performance applications also make FPGAs ideal for use in medical applications. Along with the high performance that comes from their parallel programmable logic resource, FPGAs provide a deterministic and low latency response. They also enable the implementation of safety and security structures, such as high-reliability state machines, triple modular redundant implementations, and the use of single-error correction and double-error detection (SECDED) codes on internal memories. These are user space features combined with features provided in the device itself, like configuration correction, secure configuration, and configuration memory protection. It should be noted, however, that while FPGAs are an integral part of the medical equipment, developing medical equipment that successfully passes the necessary certification requires a systems engineering-based approach that follows a well-defined process.

FPGAs in Medical Imaging

One of the major use cases of medical FPGAs is medical imaging, including 3D imaging systems such as magnetic resonance imaging (MRI) or computed tomography (CT) and 2D imaging systems like ultrasound.

3D MRI technology requires significant processing, both during the scan and after. A scan consists of two elements: the scan, during which the data is acquired, followed by the reconstruction. During the scan, the data samples are captured along a pre-defined trajectory. These samples are spatial in nature and are in what is called the k-space domain. Transforming the acquired samples into an understandable image occurs in the reconstruction phase. MRIs face the competing challenges of generating high-definition imaging, maintaining low signal-to-noise ratio, and performing fast scan times.

The complexity of the image reconstruction depends upon the sampling trajectory. A simple Cartesian scan direction aligns the k-space samples to a grid, allowing quick image reconstruction using a fast Fourier transform (FFT). A non-Cartesian scan, like a spiral trajectory, for example, results in the k-space samples being aligned in a more complex pattern, which requires advanced image reconstruction algorithms.

These algorithms can leverage the parallel nature of the FPGA logic to implement the multiple FFTs that are necessary, along with other reconstruction algorithms. To accelerate the development of these algorithms, they can be implemented using high-level frameworks such as Vitis HLS, Vitis Model Composer, or MATLAB HDL coder. Such high-level frameworks can enable developers to focus on the algorithm to be implemented and not the underlying implementation. This high-level approach can enable shorter development times.

FPGAs in Robotic Surgery

While medical imaging plays a critical role in the medical field, one of the more groundbreaking aspects of medical technology is robotic surgery. In robotic surgery, the robot and its actuators are controlled and supervised by the surgeon. Since the robot interacts physically with the patient by using actuators designed to cut and manipulate human tissue, extreme care must be taken during its development to prevent any failure that could negatively impact patient outcomes.

Successful robotic surgery requires a combination of high-performance and low-latency imaging to enable the robot and surgeon to see the surgical area clearly. To reduce the incision size, surgical robots need dexterity, which requires precise control of actuators and end effectors. This means the motor drive system must be accurate, smooth, and without jitter. The position reporting of the actuator must also be very accurately determined, requiring high resolution.

When it comes to implementing the image processing system required for robotic surgery, FPGAs offer a great solution. Their programmable logic enables the implementation of an image processing chain that is low latency and deterministic, which is exactly what is needed when guiding robotic actuators. FPGAs also make it possible to implement multiple pipelines in parallel to support multiple cameras if desired, further enhancing situational awareness for the robot and surgeon.

Programmable logic also allows multiple motor positioning and drive algorithms to be implemented, and with multiple elements implemented in parallel, the system's responsiveness and determinism further increases.

With FPGAs properly developed and implemented in medical technology, surgical robots improve patient outcomes by minimizing incision sizes and the invasiveness of procedures. This leads to a reduction in complications and creates faster recovery periods, with patients experiencing shorter hospital stays, reduced postoperative pain, lower risk of infections, and enhanced cosmetic outcomes.

It is these benefits that have made robot-assisted surgery one of the fastest growing disciplines within clinical surgery and healthcare.

Conclusion

Medical imaging is a key technology that has allowed medical professionals to understand the human body in greater detail and enable advancements in treatments. Additionally, robotic surgery has significantly improved patient outcomes thanks to its ability to enable less invasive operations. FPGAs play a central part in enabling and advancing both these medical technologies and creating better healthcare.