Rising Fortunes For ICs In Health Care

Semiconductors are increasingly finding their way into a variety of medical devices, after years of slow growth and largely consumer electronics types of applications.

Nearly every major chipmaker has a toehold in health care these days, and many are starting to look beyond wearable such as the Apple Watch to devices that can be relied on for accuracy and reliability. Unlike in the past, these chips also are being developed at relatively advanced processes, several generations behind the leading edge so that the processes have matured sufficiently.

Included in this mix are custom ASICs, as well as off-the-shelf analog and digital chips. And they span everything from low-power ICs for medical implants to high-performance accelerators that process images for diagnosis.

“Right now, chips manufactured at 28, 20, and 16nm are in production within medical equipment,” said Subh Bhattacharya, Health care & Sciences lead at Xilinx. That can include anything from a 510(k) device, which is demonstrated to be safe and effective, to De Novo and pre-market approval (PMA) classifications, which are further along. Many of these applications are Class II, which means they have a moderate to high risk for the user, but there are also some Class III applications, such as automated external defibrillators and infusion pumps.

As of today, the main uses of electronic medical devices are for research, collecting of public-health data, maintaining and accessing patient records, and patient health care. But the use of semiconductors in these devices is widening as devices shift into monitoring, diagnosis, and treatment via surgery or other therapeutics, such as drug delivery and stimulation of nerves in neurotechnology. Monitoring systems also include implanted clinical devices used at the patient’s home to send data back to doctors or monitoring services about vital-signs systems from sensors physically applied to the patient.

The over-arching goal in medical devices is always to make health care accurate, effective, convenient, accessible, and affordable, while making treatments safer and less invasive. Those improvements focus on using less power in a smaller form factor, and getting the signals out securely to their destinations. Sensors, analog-to-digital converters, RF, and microcontrollers are all key elements. So are image and signal processing. Increasingly, AI is being added into these systems, as well, to make monitoring and diagnosis tasks less complex, faster, and hopefully more accurate than humans.

Device form factors
Medical devices for patient health care take many forms, from tiny ingestible and implantable devices to large biomedical diagnosis and treatment machines. Devices for the consumer market are distinct from clinical/professional devices in that they may cost less and be less accurate or reliable.

What’s less obvious is the fine line between them. Devices used in clinical settings can have similar but better quality components and system design. But the lines are blurring as chips are used in more devices, including:

• Externally worn devices. These include wearables, hearables (fitness and vital sign trackers; newer consumer watch/smartphone systems offer ECG tracking, sleep apnea detection, arrhythmia detection, and blood oxygen tracking); disposable sensors (for example, in glucose monitoring); and e-skin for stimulating nerves, adding a sense of touch in prosthetic hands. Some of these devices are consumer devices (over the counter — anyone can buy the device without a doctor’s prescription) or clinical (prescribed by medical professionals).
• In-body devices. This group includes implants, such as pacemakers, cochlear implants, and neurostimulators. These devices have to be in hermetically sealed packages, which may be ceramic or metal. The wireless communication and signal acquisition and processing has to be extremely low-power. Electromagnetic interference (EMI) is a hazard to implanted devices. Also in this group are ingestibles, which have to be the size of pill or capsule, and made of biocompatible materials that can withstand the GI tract while housing MCUs, sensors, memory, and power supplies. Some ingestible sensors can communicate in real time with the outside world.
• Clinical patient diagnosis, care, and treatment. These are predominantly monitoring systems, including sensors connected to in-hospital networks that monitor patients’ vitals. Also in this class are lifesaving machines, such as ventilators and defribrillators, surgical equipment, robotics, and diagnostic equipment (CT scans, MRIs, X-rays). Some clinical devices are designed to be used by patients at home and some have an implantable or wearable version. Blood analyzers and assays for diagnosis are progressing.
• On top of these are some new applications, as well.

“Also becoming very interesting are assays, in which you’re in the field and want to see whether a blood sample has a certain pathogen or not,” said Aveek Sarkar, vice president of engineering for Synopsys’ Custom Design Group. “How do you quickly diagnose this? People have an image sensor kind of device where a lot of the pixels are captured with the light in your phone. These are biomedical sensing devices that essentially have sensors or markers that can study very, very quickly, what are the components that are present in the sample that you have collected,”

Virtually every type of chip, from MEMS to DSPs and FPGAs, is being used for medical applications. Some of these are planar chips, some are in packages, and increasingly they have built-in security. But with the mashup of ICs in medical devices, often the answer to some medical device design problems is a custom ASIC, which can be designed to fit an exact system and function.

“The majority of the ICs are ASICs, specifically customized for meeting the customer’s requirements,” said Ralf Leuchter, segment head of Customized Solutions at Infineon Technologies. “Typical applications include X-ray counting IC for computer tomography, hearing aids (digital-ICs with µC, DSP, and audio engine; mixed signal ICs with support function like ADCs, DACs, microphone-IF, supply, etc.), continuous temperature measurement with NFC read-out (for patient monitoring, pregnancy control, and transportation of medical goods, etc.), and blood analysis (blood sugar, coagulation, etc.). In addition to these applications, standard ICs are commonly used in AC or DC supply solutions along all typical medical applications.”

ASICs simplify the regulatory side of things. “As ASIC devices are tailored to customers’ specifications, there are normally no surprises during the certification,” Leuchter said. “Hardware (ICs) can have a huge impact on passing the regulatory, but when they are designed together with the entire system, this risk is nearly eliminated.”

But ASICs have limitations, as well. “Anything that is expensive takes a long time and it’s risky, so don’t do it unless you’re basically stuck,” said Andrew Kelly, director of applications engineering at Cirtec Medical, a company that designs medical devices. “For a lot of projects we look at, we could say, ‘Technically, here’s a beautiful ASIC and it’s wonderful solution for you, and isn’t this great?’ But financially, it just doesn’t add up. It’s very common.”

FPGAs
Where change is rapid, the risk of obsolence can be problematic. This is why FPGAs and SoCs with some level of programmability are widely used in medical equipment and applications.

“Some of the key attributes that make FPGAs ideal are low latency or real-time requirements, such as for surgical applications or endoscopy pre-processing,” said Xilinx’s Bhattacharya. “They also may have, for example, a high rate of data movement for beamforming in ultrasound or back-projection in CT scanners, image reconstruction algorithms, etc. Additionally, FPGAs offer some unique capabilities like flexibility and future-proofing of IP, easier integration with analog front ends (AFEs), and better safety features.”

Bhattacharya noted that FPGAs and adaptive SoCs are used in a variety of modalities, including medical ultrasound, digital X-RAYS, CT, MR and PET scanners, and in diagnostic, surgical and other clinical equipment. “Areas like smart beds, 3D dental imaging scanners, multi-parameter patient monitors, AEDs, and defibrillators are some of the fastest growing medical applications for us with our SoCs.”

Some areas where he said FPGAs excel, include:

• Signal acquisition, capturing, management of sensor data from one or many sensors;
• Complex digital signal processing tasks in beamforming in ultrasound, back-projection in CT scanners, 2D-FFT (Fast Fourier Transform) in MRI, complex math functions, etc.;
• Raw video pre-processing in endoscopy systems and video post processing, and
• Real-time response, which is particularly interesting because in CPUs or GPUs, software will periodically poll memory, causing interrupt delays. One example is real-time alarms for nursing stations on patient monitors.

Photonics
The role of light in medical applications is both logical and practical, providing extremely high performance with low heat. Rockley Photonics supplies a digital sensor for the Apple Watch. “It’s a full stack — what they call a clinical wrist digital sensor system and it’s a photonic IC,” said Synopsys’ Sarkar. “They design their systems on our platform.”

Rockley is working on non-invasive biomarker sensing that uses laser, rather than LEDs, to detect vital signs. “They are using infrared spectral photometers, for example, that measure light. Historically when you look at photonic devices, they came out into the world for long-haul communication, because by using light we could have a much faster and also sometimes lower power waves. The application for photonic devices is getting extended into lidars, and now also biomedical applications like this kind of sensor system,” said Sarkar.

Synopsys worked on early photonics projects and saw its use broaden. “Typically when we created the software, the design environment was used by our partners, which are designing long-haul communication platforms like for a high-performance compute rack,” he said. “I’m sending large volume of data between one rack, one blade to another, and I need to send them very, very fast. That kind of a very, very high-end environment was what a photonic IC was used for. And so they will use our entire design platform. But then, when the Rockleys of the world start to take some of these technologies, and then drive it for these biosensing applications, they see the work that has been done there and accelerate their design process using these technologies.”

3D ICs
Stacked dies is showing up more frequently in medical, as well, both due to space limitations as well as the ability to cram more functionality into a given area. “The medical domain is a target and technology driver for 3D ICs as they have advanced requirements for form factors, power, energy, thermal, and electromagnetic effects,” said Frank Schirrmeister, senior group director, Solutions & Ecosystem at Cadence. “The overall market size is not as large as what is going on in mobile and hyperscalers, but is growing with populations getting older. And they do drive advanced technology requirements.”

A 3D-IC integrates different processing, memory, RF, sensors, and other blocks onto multiple stacked silicon chips connected by some type of connector, such as through-silicon vias. The chip saves space, but its complexity makes for tougher design and integration issues.

MMICs
The microwave monolithic integrated circuit (MMIC) is an example of an IC type being adapted for a medical use. The MMIC is being used to sense temperature on cardiac catheters. The AWR division in Cadence worked with Meridian Medical Systems (MMS), to develop a catheter for cardiac arrhythmia procedures that simultaneously delivers microwave radiation for tissue heating and a radiometer (essentially a remote sensing device), which was fabricated as an MMIC to sense the temperature of the heart wall. The doctors doing the procedure can see in real time if there are temperature drops.

Anti-tamper ICs
Medical device designers are becoming more concerned about protecting their brands from re-purposed or cloned medical devices. Any throw-away medical device, such as a disposable sensor, could be retrieved, repackaged, and resold, or the whole device or part of the device can be cloned. “In general they’re reluctant to share a use-case failure because of the implications to the brand,” said Scott Best, director of anti-tamper technologies at Rambus. But customers do report that tampering is a principal concern, because it can affect patient safety and tarnish a brand.

The medical device industry has learned from the printer and camera industries, which aggressively protect their consumers against counterfeit products. Canon, for example, put anti-counterfeiting technology on its lithium ion batteries and battery chargers for its SLR cameras to make it difficult to use counterfeit chargers that cause batteries to blow up.

“When the medical industry started having access to somewhat portable electronics, like battery-powered electronic monitoring devices that could be attached to users — that’s somewhat a more nascent market, of course, then digital SLRs — the medical industry could definitely look and say, ‘Alright we’re going to protect our revenue, we’re going to protect our brand, and we’re going to lead by saying we’re protecting patient safety,’” Best said. “You can accomplish all three, and you can do it for both the best of reasons and for the most financial reasons — all at the same time.”

A cloudy future
Medical devices have been communicating data to doctors for some time, allowing them to make more informed decisions more easily. “Clearly there’s a lot of technology in medical equipment, especially imaging equipment, and so much of the DSP technology and signal processing technology has found a good home in medical,” said Sam Fuller, senior director of marketing for AI Inferencing at Flex Logix. “But those have mostly been around capturing images digitally and being able to store them. Now, the advent of AI allows for the understanding of the image.”

That offloads the task from radiologists, allowing them to be more productive. The data then moves somewhere else for processing, or to an edge device.

Edge devices and cloud connections in medical devices are still in early days, but they are here to stay. “The challenges in cloud connected devices are causing a revolution in the way edge devices are developed, and this brings significant benefit for medical device development,” said Peter Ferguson, director of health care technologies at Arm. “By improving the synergy between the device and cloud software, the silos that have been created by technology differences and lack of systemwide standardization can be better integrated. Initiatives like Arm’s Project Cassini, a standards-based initiative to deliver a cloud-native software experience at the edge, are helping to address this.”

Still, much work lies ahead. “The industry must work together to address these challenges if we are to reach the next phase of growth in digital patient care and continue to enable breakthrough technologies that can have a huge impact in health care and medical applications on a global scale,” said Ferguson.

Conclusion
Medical chips include a range of chips and innovations used by the medical industry to improve health care. The research is ongoing, and new uses and technology are being developed all over the world. The current pandemic has accelerated some of these innovations, but there are many more to come.

Taken as a whole, these will open broad new business opportunities for the chip industry. After years of frustration in moving this market forward, the logjam finally seems to be breaking open.

Related
Overview Of Medical Chip Challenges
What challenges medical device designers face, and how EDA helps.