How to Choose an Edge AI Computer for a Medical Device

By Mike Pizzirusso, Head of Product Development, FORTEC US

Choosing an edge AI computer for a medical device comes down to three things: whether it runs your validated AI workload inside the device’s thermal and power limits, whether it will stay available for the life of the product, and how cleanly it integrates with the display and power system around it. A headline TOPS figure is only the starting point. FORTEC US supports five of the world’s top ten medical device manufacturers with display, embedded computing and power technologies, and the decisions below are the ones we see shape every program.

What is an NPU, and why does it matter in a medical device?

An NPU, or neural processing unit, is a processor built specifically to run AI models. Intel Core Ultra processors combine a CPU, a GPU and an NPU on one chip, so AI inference can run on dedicated hardware instead of competing with everything else the device is doing.

That matters more in a medical device than in a typical PC. A patient monitor or diagnostic instrument has to keep its user interface responsive, process sensor or image data, and log results at the same time. Moving sustained inference onto the NPU leaves the CPU and GPU free for that work. It also helps with heat, because the NPU is designed to run AI tasks efficiently at low power. In a sealed or fanless enclosure, where airflow is restricted, the thermal budget is often the real limit on how much AI a device can run.

What does “up to 100 TOPS” actually tell you?

TOPS stands for trillions of operations per second. It is a peak figure: the most AI arithmetic a platform can perform under ideal conditions, and a headline TOPS number is usually the combined capability of every processing engine on the chip. It is useful for comparing platforms. It is not a measure of how your own model will perform.

The right architecture depends on factors a datasheet cannot settle on its own:

• The validated software workload the device will actually run
• The performance target for that workload
• The thermal design of the enclosure
• The lifecycle requirements of the program
• The cybersecurity strategy for the device
• The regulatory responsibilities of the medical device manufacturer

Treat TOPS as the ceiling, then confirm the real workload runs within the device’s power and thermal envelope.

Which medical applications benefit from edge AI?

Edge AI means running AI models on the device itself rather than sending data to the cloud. Processing data locally reduces dependence on cloud connectivity and supports responsive, data-intensive applications close to the point of care.

AI-capable embedded platforms can support:

• Image and video analysis
• Diagnostic workflow assistance
• Smart laboratory automation
• Predictive maintenance of the equipment itself
• Advanced HMI functions
• Real-time data processing

These are the workloads driving demand across medical imaging, diagnostics, workflow optimization and equipment monitoring.

What are the iBASE MEC100 and MED3100?

The MEC100 Medical Edge AI Computer and the MED3100 Mini-ITX motherboard are recent iBASE introductions for AI-enabled medical equipment. Both are based on Intel Core Ultra Series 3 processors and are specified for AI performance up to 100 TOPS.

They suit different stages of integration. The MEC100 is a complete medical AI computer, for OEMs who want a finished edge AI system to build around. The MED3100 is a Mini-ITX motherboard, the standard compact board format, for OEMs who are designing the computing platform into their own enclosure.

Why should the computer, display and power be specified together?

Because in a medical device they operate as one system. The display, the embedded computer and the power supply have to work together as a reliable architecture that supports today’s software and leaves room for future development. Choosing each part separately, against its own specification sheet, is how integration problems end up discovered late in a project.

For many devices, the display and touch interface are what clinicians judge the whole product by. FORTEC supplies TFT displays in a wide range of sizes and formats, and for custom projects develops complete display assemblies that combine projected capacitive (PCAP) touch, custom cover glass, optical bonding, controller integration and mechanical design. Bonding the cover glass to the panel, tuning the touch controller for gloves and disinfectant, and designing the mechanics around the enclosure are engineering decisions, not extras. Each one affects how readable, durable and easy to use the screen is. Our own VacuBond® optical bonding technology is part of that work.

Behind the display sit the embedded computing platform and the AC/DC and DC/DC power supplies that feed it. Evaluating all three together, from the earliest design stage, reduces integration effort and technical risk, and keeps the platform flexible for future software and hardware generations.

How do you plan for a medical device’s long service life?

Start at the concept stage. A medical device is expected to ship, and be serviced, largely unchanged for a long time after launch. That makes the availability of every component, consistency from one production batch to the next, and access to technical support as important as the processor or the resolution of the display.

Parts with a published roadmap and a stable supply chain mean fewer forced redesigns, fewer interruptions to the qualification work a medical program depends on, and a lower cost of ownership overall. Weighing availability at the same stage as performance, rather than after it, is what lets one platform carry production, field service and the next product generation. Our guide to component obsolescence covers how to plan for it before it becomes a problem.

Why work with FORTEC US on a medical device program?

FORTEC US currently serves five of the world’s top ten medical device manufacturers, on programs that depend on reliable display, embedded computing and power technologies. That experience covers the practical challenges medical OEMs face: system integration, component lifecycle management, documentation, supply continuity, and the growing need to support AI-enabled workflows.

Our engineers work alongside medical OEMs, design houses and system integrators from the first concept to volume production, with US-based technical support. Understanding the clinical setting, the people using the device and where each program needs to be in several years is how we help customers find the balance between performance, reliability, ease of integration and lifespan that suits them.

FORTEC US at a glance

• Serving 5 of the top 10 global medical device manufacturers
• Custom medical display and HMI solutions
• AI-ready embedded computing, including the iBASE MEC100 Medical Edge AI Computer and MED3100 Mini-ITX motherboard
• AC/DC and DC/DC power solutions for medical system architectures
• Lifecycle, integration and US-based technical support

Planning an AI-enabled medical device? Speak to our team to discuss your specification.