Part of the  

Chip Design Magazine


About  |  Contact



From MEMS to COMs: New Technologies Help Shrink Medical Devices

By Cheryl Coupe

Embedded developers often talk about scalability and the design challenges of supporting a broad range of performance and cost options. But medical devices may provide some of the most dramatic examples of scalable designs. Look at medical imaging devices. At one extreme are million-dollar-plus magnetic imaging resonance (MRI) machines. For this typically room-sized equipment, the latest market opportunity may be for even larger, more powerful devices that can accommodate America’s growing number of obese patients and penetrate larger tissue mass. At the other extreme are swallowable capsules that include a tiny video camera, lights, transmitter and batteries to provide wireless scans of the small bowel while a patient carries on close-to-normal activities at home. Our experts discuss the evolutions and technologies that are driving both ends of this scale. With us are Steve Kennelly, senior manager, Medical Products Group, Microchip Technology Inc.; Dan Demers, sales and marketing manager, congatec; Cameron Swen, strategic marketing manager, AMD Embedded; and Matthias Huber, vice president of marketing for N. America, ADLINK.

EECatalog: What are some of the challenges and opportunities being driven by the advent of wearable, wireless, extremely low-power medical and fitness devices?

Steve Kennelly, Microchip Technology Inc.: The technical challenges of wearable devices are pretty well understood—on one side of the equation there is functionality and features and on the other side there is the size, weight, flexibility and cost. Many levels of feasibility have already been demonstrated with wearable glucose monitors, insulin pumps, acceleration-based pedometers/activity monitors, heart rate monitors, GPS trackers, etc. I think the next big challenge looming on the horizon is related to marketing and user acceptance. It remains to be seen how many things people will be willing to carry or wear, and what minimum number of such things will be required to measure everything that people want to measure.

Dan Demers, congatec: The challenges, as found in other vertical markets, include the ability to design standard platforms that are small enough and low-power enough to fit in the needed envelope. In addition there is the challenge of making sure that the platform as a whole addresses shock and vibration. As silicon vendors continue to shrink their processes, the board and platform providers follow suit. We are now witnessing the entry of the SOC in x86 processors. This is another milestone for embedded computing in general. The SOC approach is driving down size and power even further. This is creating vast opportunities for board and system builders when it comes to designing the next generation of small, mobile, low-power devices.


Matthias Huber, ADLINK: Since ADLINK is not designing these devices (also called gadgets) I can only answer from our perspective. Those devices generate data 24/7; this data may be consolidated by a phone for an individual at home. In a managed-care or hospital setting this is more complex. The data aggregator must be trusted and secured, meaning the patient data can only be used if the chain of trust is intact. So there is opportunity in dedicated devices that aggregate data, provide a user interface, and even management of the gadget itself. Some of those points are in the home health segment, assisted-living or hospital settings

EECatalog: Dramatic innovations in MEMS and sensors have revolutionized consumer electronics. What are you most excited about in terms of how these technologies will change medical device development? What challenges are you seeing in the development process with these technologies?


Kennelly, Microchip Technology: The most exciting sensor-based applications I see coming are in the field of in-vitro diagnostics. Commercialization is still a ways off, but I am intrigued at the progress being made in using MEMS and other technology to create fast point-of-use diagnostic devices for evaluating properties of blood and other fluids. A low-cost and fast test for specific pathogens would be invaluable for preventing future viral outbreaks and reducing the spread of antibiotic-resistant bacteria strains. Like everything, the challenge is getting the cost down low enough so that the technology becomes cheap enough for mass distribution. The historical cost trends of other MEMS applications give some encouragement in this area.


Demers, congatec: There is often the “5 pounds in a 1-pound bag” scenario occurring with system solutions. There are so many opportunities to highly integrate a system-level solution today. Many aspects can be found native in the silicon unlike a decade ago. It seems as though the list of “nice-to-haves” continues to increase, but in reality, there is only so much space available when the final product plan still emphasizes a small form factor. That being said, there will continue to be this push towards connectivity. It is seen in the medical equipment world and many other industries as well. This is very exciting as it opens up many new opportunities for what these devices can offer to the medical community.

Huber, ADLINK: We all have seen the evolution of consumer devices in the last few years. Small medical devices may change to relying more on having an external computing element, and data may be transmitted using Bluetooth 4.0 or other short-range radio. So for the designer of a device, power management and radio technology will be much more relevant. For dedicated devices (mainly professional use in GP or hospital) similar rules apply. The challenge will be in the increased cadence of innovation; however, the regulatory bodies are the ones setting some of the time horizons.

EECatalog: How are designers keeping up with need for improved graphics in everything from imaging applications to robotic surgery to telehealth equipment?

Kennelly, Microchip Technology: Ever-increasing graphics capability is an area where medical device design is fortunate to be a technology follower. Gaming systems, television, mobile phones and other personal electronic devices will always push the need for improved graphics faster than medical devices. The challenge for medical device designers is picking up the appropriate technology for the device that they are designing. With so much development happening in graphics and display technology, it is important to be selective and consider things such as product lifetime and environmental compatibility.

Demers, congatec: There is no doubt that graphics performance is a large aspect of embedded computing design these days. All silicon vendors are finding ways to increase this performance. There is still the option to use external graphics controllers; however, the native performance is really making a lot of headway. This is helping to once again keep the size, weight and power draw down in the system solution.


Cameron Swen, AMD Embedded: If we are talking strictly graphics, supporting beyond HD resolutions is essential in these applications. When a doctor is analyzing a diagnostic image it needs to be delivered on a very high resolution such that pixilation of the display doesn’t affect what he sees. But when talking about applications such as medical imaging and robotic surgery, it is about much more than assisting the physician to see the surgical area. The image processing capabilities afforded by the integrated graphics in an APU or a discrete graphics processor can help to enable higher precision and speed for robotic or surgical assist devices to improve stability and reliability, as well as helping enable image enhancement technology such as being able to see through fluids or tissue.

Huber, ADLINK: This will vary quite a bit depending on the use case. A control application with a “static” user interface will likely be served with the capabilities of current SOCs; however the expectations on the (touch) GUI pose a demand on the video processing that may consume more compute power than expected. In imaging, more is always better. What we see is that the algorithms are tuned to use up all available compute performance. The effect may be that a mid-end ultrasound device may use on-chip video processing instead of an external card, while those will still be used in the high end versions. Applications like telehealth will need good video streaming (encode/decode) capabilities, however in many processors we find those tasks solved in hardware-based solutions. For a straightforward video conference that is not a big challenge; when we look at diagnostic imaging or surgical imaging, this becomes much more demanding as the image quality losses (artifacts) of the compression algorithms are not tolerated. Robotic surgery is a very interesting area; a surgeon can be “inside” the patient using 3D camera and viewing systems. These imaging systems require quality even above broadcast standards, HD resolution, 60FPS, 30 bit color, stereo and no losses or artifacts. These demands are often met with a high end capture card with a dedicated hardware or GPGPU solution. One last comment: due to the massive data volumes generated this cannot be done remotely due to issues of latency.

EECatalog: Which board and computer-on-module (COM) standards are seeing the greatest success in the medical market? In your view, what still needs to happen within those standards to enable their success?

Demers, congatec: When it comes to handheld, low-power devices, the Qseven specification supported by SGET is really at the core of COM standards. Released in 2008, it has the advantage of being first-to-market with an edge-connector approach and an emphasis on highly integrated, small form factor, low-power computing, especially for devices that demand mobility. For continued success, the Qseven standard will need to continue to follow the silicon trends of smaller, lower power, etc. (additional smaller form factors for Qseven are already adopted by SGET). In addition, like all standards, Qseven will need to keep up with newer widely adopted buses and I/O. COM Express is a great example of how industry standards can keep up with silicon technology advancement. Initiated in 2004, it is still one of the most scalable and widely adopted COM standards. When it comes to devices that really need higher end graphics and processing, COM Express is most often the go-to for designers.

Swen, AMD: From what we have seen, the COM Express standard seems to be very popular in medical applications. The size of the COM Express basic form factor is large enough to easily dissipate the thermals in a design that supports both a high-performance APU and a discrete graphics processor on a single module. Combined, the APU and graphics processor can deliver over 1 TFLOP of compute performance for applications such as ultrasound computed tomography, while balancing the power consumption to enable applications to be portable. And many of our customers are finding homes for their smaller AMD G-Series-based COM Express compact modules in small form factor applications such as bedside terminals and patient infotainment that require low power but still must deliver very good graphics performance. While COM modules are intended to simplify the design for an OEM, the key functionality that seems to be winning designs in these applications is providing the features and support to genuinely make integrating them into a baseboard easier to shorten the time for an OEM to design and bring up their first systems. And offering processing solutions with consistent features and software support across price-, power- and performance-scalable modules makes it easier for OEMs to leverage their expensive software investments across their product lines.

Huber, ADLINK: The most successful standard in COM is the COM Express standard. It has seen the widest adoption so far. The main reason for this is the time of introduction, the fact that it is a real multi-vendor (no de-facto) standard held by an independent specification body. Now recently a few new standards are coming into the game like Qseven. While it cannot address all performance levels, it is more cost-effective and allows for a thinner profile.

The latest entry to the embedded standards for COM would be the SMARC format. Initially geared to embrace the ARM-SOC interfaces, the up-and-coming generations of Intel ATOM will also make a good fit. The standards have to live up to a few expectations. For one, they must be relevant; meaning size, power and performance must match up and be beneficial in the use case context. They have to solve a real-world problem in an effective and preferably innovative way; usually this would be time-to-market and risk reduction. Last but not least, the standards must be truly a multi-vendor standard with an independent spec organization.

EECatalog: Software has become an increasingly important and complex element of medical device development. How is this changing your business model?

Kennelly, Microchip Technology:Software is at the heart of any embedded design, including medical devices. It is the software that puts a designer’s intellectual property into a standard MCU and drives the differentiation of the end device. Unlike in many other designs, medical device software can have a direct impact on the safety of the user. We are seeing an increasing number of customers asking about safety certifications of our tools, including C-compilers. We are responding by ensuring that our development tools comply with applicable safety standards. We have also published a functional safety manual to assist software designers in meeting their system software compliance requirements. These resources have enabled Microchip’s customers to achieve third-party confirmation of their system’s compliance with IEC 61508 and other functional safety standards.

Demers, congatec:When offering a standards platform, it is important to make sure there is support for the many operating systems in use today. We have seen new operating systems enter the market in recent years, i.e., Android. In addition, there is an increase in ARM technology that is disrupting the historical x86 platforms. How the operating systems and software engage these hardware platforms is of great consideration. Making sure that all of these bases are covered has changed the way many embedded computer companies approach the market. We are seeing more technology partnerships develop and we are seeing support and engineering staffs broaden in expertise. Many embedded companies are creating focus product and support for the medical market. Overall, there is certainly a more strategic focus being employed to specific vertical markets like never before.

Swen, AMD:With their high development costs and long design cycles, medical imaging applications lend themselves to taking advantage of innovative solutions. Designers of medical imaging applications were some of the first pioneers to begin taking advantage of stream processing on graphics processing units. This has had a big impact on AMD’s vision of silicon and software architecture as stream processing evolved into OpenCL today. Medical applications are one of the fastest applications adopting and innovating with OpenCL, where they are taking advantage of its performance capabilities as well as software re-use across platforms to easily scale up and down between very high-performance fixed installations and low-power portable applications. These software architectures are continuing to evolve into the Heterogeneous System Architecture (HSA); an open, standards-based approach to heterogeneous computing that will provide a common hardware specification and broad support ecosystem to make it easier for software developers to deliver innovative applications that can take greater advantage of modern processors from a variety of vendors.

Huber, ADLINK: We have to look at what part of the software stack we look at. Usually board-management tools are vendor-provided and free. The needs for security and manageability are met by integrating solutions from ecosystem partners, and the business model would include the license handling.

Looking forward, some of the software features will be enabled in the field and cloud-based solutions may have a subscription or volume use pricing model. None of those business models is new in itself; it is, however, new to embedded or medical device makers.


Cheryl Berglund Coupé is editor of EECatalog. com. Her articles have appeared in EE Times, Electronic Business, Microsoft Embedded Review and Windows Developer’s Journal and she has developed presentations for the Embedded Systems Conference and ICSPAT. She has held a variety of production, technical marketing and writing positions within technology companies and agencies in the Northwest.

Tags: , , , , ,

Leave a Reply

Extension Media websites place cookies on your device to give you the best user experience. By using our websites, you agree to placement of these cookies and to our Privacy Policy. Please click here to accept.