Imaging systems, particularly internal-imaging systems, play a key role in modern medicine. Once an image is acquired, it can be used to diagnose or decide the nature of intervention required. For example, an X-ray of a broken limb can help determine the extent and nature of bone injury (diagnosis) while an MRI of a tumor can help decide the details of the intervention (surgery) required, like where and how deep.
|
You acquire such images of human internals using X-rays, radio frequencies or magnetic resonance (MR) depending on the organ being investigated and what the end use of the image is to be.
Once an image has been acquired using a scanner or camera, the rest of the work on it can be done using software, at a
viewing station.
Unix & Windows
Unlike commercial PCs or servers, medical-imaging systems have a very long lifetime (ten to twenty years and more). Also, hardware and software upgrades are not that frequent or easy as with traditional computing systems. The software for medical-imaging systems run on systems that have traditionally run different flavors of Unix, like Solaris and even Digital VAX. More recently, developers have started experimenting with the embedded versions of Windows.
From the traditional monolithic architecture of the software and hardware, there is currently a move towards using off the shelf components and componentized software.
Standards
All software currently being written for medical-imaging systems have to confirm to the DICOM (Digital Imaging in Communication in Medicine) standards to ensure that different systems from different vendors can successfully share information. So, you can, for example, acquire the image from a Siemens, and do the processing on a Philips viewing station.
Viewing stations
Multimodal viewing stations (the same station being able to process say MRI as well as CAT scan images) are already in common use. Vendors are also able to send private information that only their own software and viewing stations can read, so as to enhance their equipment. For example, a Philips image-acquisition system can acquire and transmit more information than prescribed by the standard. Such extra information can be deciphered only by a compatible Philips station, while say a Toshiba viewing station would get only the information prescribed in the standard.
|
Even though the basic job is that of image processing, the algorithms used in medical software can be vastly different from say those used in other commercial image-manipulation software like movie software or Photoshop. The reason behind this is that the medical systems have to preserve a very, very high degree of accuracy and detail, or there could be fatal results, while such constraints may not exist for commercial image-manipulation software.
Color or grayscale
Overall, the use of color is less prevalent than the use of grayscale in medical imaging. But, there is a slow changeover to color where it adds value. For example, in MRI, color can better depict variations in internal tissue layers, and hence using color here can add value. On the other hand, X-rays just provide a two-dimensional image of the bone structure, and hence adding color here may not be of any value. To get color, the acquisition system need not be color enabled. The processing software can subsequently add color to the image.
The Internet has caused another major change that is happening in this field. On one side the equipment needs to be made capable of being used for remote diagnostics and tele medicine applications. On the other side, Internet enablement makes it possible to remotely diagnose, repair and even software upgrade the machines themselves!
How much effort does it take to create such a system? It takes up to a thousand people years (a hundred to a hundred and fifty people working for about five years) to develop an image-acquisition and processing system, ground up. And the bulk of the work is in the software effort.
Shiva Kumar KR, is Lead Software Architect at Philips Medical Systems, Bangalore