By Garrett Seeley
I recently went to an OEM training for an imaging system. To say the least, it has been an eye-opening experience for me, especially in application in networking. Usually, networking in medical devices means a communication over the hospital network. That is specifically, when working on room sized imaging systems, there are networks outside of the room that we communicate to. This is nothing new. What I found that is new is that networks may be used inside the equipment itself. Call me old school, but this was a revolution in thought for me. It was surprising to me to see the varied networks transferring information between sub-assemblies, gantries, and host control systems. After my training, I found similar networks used not outside of the medical devices, but inside of them as well. It is a revolution, or perhaps an evolution, in medical devices and it is already in full implementation. I want to elaborate how TCP/IP, Bus, and Ring networks that are enhancing and transforming the internal design of medical devices.
For example, recon imaging and hosts systems typically use star networks and TCP/IP to communicate. It is the most widely used network format in a hospital setting. It typically connects devices to EMR or PACS systems. We have covered TCP/IP quite extensively; however, TCP/IP is mostly used for external communications to a larger network. Inside the machines themselves, it is mainly used for computer-to-computer communication. Think of a recon computer. It needs information from an imaging host (main control) computer to perform post processing of images. Recon is usually done on a stand-alone computer system networked to the host, but not networked to the outside DICOM network. This is usually performed on a TCP/IP format and does not have to connect to the main hospital network. The same network can exist in image acquisition. The scanning system may be a stand-alone computer on its own, separate IP network tied to only the host system. Inside the imaging system, between acquiring the image and displaying the image at the control, a TCP/IP network may be used to transfer the image between a detector and a control computer. Although this is an example of TCP/IP used inside what most people call one medical imaging system, it is not what I am going to focus on for this article. If you need more information on TCP/IP, please refer to the previous Networking Notes articles. This is a clear use of networking for a medical device running more than one computer for a room-sized equipment. However, let’s move on to what happens more under the covers of a smaller medical device.
One the most impressive use of an internal network is the CAN bus. The CAN was covered in a previous article; however, it is highly valuable to medical devices. In the older days, signals were wanalog. They only required a signal and a reference wire. Data was amplified and filtered, and received by a control circuit; well, most of the time it was received. The problem was that the signal became easily lost or corrupted because of interference. This could change measurements and result in wrong data or loss of function. As computers became smaller, the sensors themselves became smarter. Now a thermometer can have a network chip on it. That is, the sensor itself is networked. It converts the signal from analog to digital, and buses the signal on a CAN network to a controller. It still follows the same rules of an analog system, however now there are 4 wires needed – two for power and two for signal. The signal is now a digital stream at full power, reducing the risk of loss from small analog signals. This may seem like a financial waste but considering the data performance and reliability, it is well worth the cost increase. Because of this, CAN is finding its way into more and more medical devices. We are starting to see these “small networks” on devices that traditionally most would not think would need to be “smart devices,” such as warming cabinets or refrigerators. The ability to communicate via CAN to sensors or controllers in digital makes it easier for microprocessors to handle the data, eliminates filtering or amplifying the signal, and streamlines the communication of data. CAN is being used in smaller and simpler devices, upgrading their internals into small scale networks. It is the way devices are going so keep an eye out for more use of CAN in the future.
Another network I see used more often in medical devices is a Token ring communication. This has evolved from coaxial wiring to high-speed fiberoptics. In a modern system, where the data is passed from module to module, rings are a high speed, reliable option. In a ring, data is passed from components called nodes until it reaches the destination node. This is fast but becomes inefficient when dealing with more than a dozen nodes. If there is a limited number of nodes, a ring-style network ensures the data is reliably transferred as only one node is communicating at a time. All nodes are essentially equal, and special networking hardware, controller or server is not needed. The problem with this style of network is that if one of the nodes is out of operation, it shuts down the entire network. This can make troubleshooting more difficult. However, for small, fast communication between components that have the same rights, a ring is an ideal network.
Whether TCP/IP, CAN, or Ring, these techniques are being used more inside the medical devices, not just outside of the devices. Look for these elements in use on modern equipment. Do not be surprised if a brand-new warming cabinet uses a CAN network to communicate to thermometers. I can assure that the new X-ray systems may have all three of these networks inside their cabinets. It just goes to show that networking is not just for external communications, but for internal communication as well. Consider this when troubleshooting equipment.
