Hofland and van der Linden point out that magnetic resonance imaging (MRI) scanners are devices that typically stay in service for at least ten years. During that time, the manufacturer is expected to maintain the hardware and software viability of the machines. Hardware upgrades can keep an older machine viable.
As devices with computing capabilities, MRI scanners are considered low-volume devices with long life spans. Similar devices within healthcare would be the linear accelerators that are used within radiation oncology departments. Like MRI scanners, these devices are complex and require increasing amounts of computing support. It is not unusual for a linear accelerator to stay in service for 20 years if given the proper hardware and software updates.
The authors point out the daunting problems associated with maintaining MRI scanners. They compare MRI scanners with similar real-time consumer devices, such as the Bosch Engine Control Unit and the RealNetworks Helix Client. MRI scanners have an installed base of about 10,000 systems, and approximately ten million lines of code are required to support them.
This is an interesting article that points out some of the challenges associated with writing and maintaining code within a medical device environment. The authors do an excellent job by comparing a medical device (an MRI scanner) with consumer devices (an Engine Control Unit or a Real Player client).