Image is everything
Looking inside the human body has moved on since X-rays were developed with everything going digital.
Medical imaging has come a long way in the century or so since Röntgen discovered X-rays in 1895. Go to any large hospital for a scan these days and you can find yourself at the tender mercy of an imaging system using magnetic resonance imaging (MRI), a gamma camera, ultrasound, digital mammography, computed tomography (CT) or positron emission tomography (PET) - as well, of course, as X-rays.
Each system has its own advantages, and which system will be used on you will come down to whether your problem lies in an area of hard or soft tissue.
Any of us who's broken a bone or been expecting a baby will of course be familiar with X-ray and ultrasound systems. But if soft-tissue diagnostics are needed then an MRI or PET system is more likely to be used. Or they may be used in combination - PET scans can be performed on the same equipment as CT scans, so are often viewed alongside each other.
As their names imply, each technique - or modality, as they are called - is underpinned by a different physical principle.
For example, MRI uses static and varying electromagnetic fields to polarise and excite hydrogen nuclei in water molecules in human tissue and produce 2D or 3D images of the body; CT uses X-rays, also to obtain a series of 2D 'slices' through the body, and PET generates images by collecting photons produced by the decay of short-lived radioactive isotopes injected into the body. This is why PET and CT scans are often used together, as it is very useful to be able to view a scan of, say, a tumour detected by the PET system next to an image of the patient's surrounding anatomy detected by the CT scan.
So, at the imaging end atleast, the approaches to their control are necessarily very different. For example, again with MRI, the control system has to address factors such as the strength and frequency of the fields that manipulate the hydrogen nuclei to produce measurable signals and the spatial encoding to produce the images. By contrast, in X-ray modalities such as CT, the issue is inherently to do with controlling radiation exposure.
But things are different when it comes to motion control.
"Here, the issues are pretty much the same for most modalities - location of the imaging sensor, speed of travel, for example," says Srini Srinivasan, principal software engineer in the Boston office of technology design and development company Cambridge Consultants.
"With CT scanning, for example, the control systems need to deliver precise rotation and gantry motion synchronised with X-ray generation and data collection. Patient and medical personnel comfort and safety are the overriding factors in this connection."
The use of so many modalities in a typical modern hospital raises two key issues - data management and interoperability between systems.
Gone are the days of having an X-ray, waiting for the film to be developed then taking the film in with you when you see the doctor; everything is digital now. And with modern scanners producing images in unprecedented detail, the amount of digital information swirling around hospitals is huge, and growing all the time. Add to that the different types of image and data formats, and the different types of user-end hardware needed to view and manipulate those images, and you have the stuff an enterprise-wide IT manager's nightmares are made of.
So two key technologies have been developed to knit these heterogeneous systems together - PACS and DICOM.
A PACS (picture archiving and communication system) is a set of computers or networks designed to store, retrieve, distribute and present medical images generated by the various modalities, and replaces traditional hard-copy images such as film archives.
A typical PACS computer network consists of a central server with a database containing the images, which is connected to users who provide or use the images via a local or wide area network. Increasingly, a PACS also includes Web-based interfaces for communication over the Internet, usually via a Virtual Private Network or Secure Sockets Layer.
The images themselves are stored in an independent format, and the standard here is DICOM (Digital Imaging and Communications in Medicine). DICOM includes a file format definition and a network communications application protocol that uses TCP/IP to communicate between systems.
DICOM groups information into data sets so that a file of an X-ray of a chest, for example, actually contains the patient ID within the file. This prevents the image being separated from this information by mistake.
The standard also has enough latitude to allow the various imaging equipment vendors to create DICOM-compliant and non-compliant files that differ in the internal tags used to label the data and the metadata. Most PACS networks read the metadata from all the images into its central database, which allows users to retrieve all images with a common feature regardless of their originating instrument.
Ultimately this leads to a completely filmless system, although in many UK hospitals this is still some way off. One hospital that is truly filmless though is the University Hospital of Wales (UHW), in Cardiff. "We have what's called a full PACS system," explains Tom Henderson, the hospital's PACS manager, "so every modality - from endoscopy to CT and MRI - is integrated. We have our own health trust network, we're connected to every other hospital in Wales and we're also developing a secure Internet connection with a hospital in Bristol.
"For us, PACS has removed geographical and time boundaries by enabling specialist clinicians to look at patients' images from different sites at any time, cut patients' radiation doses by slashing the need for repeat scans, improved clinical workflow and freed up huge amounts of manpower and time," he says.
UHW has what's called a managed PACS service. The provider is Agfa Gevaert, which supplies and owns all the on-site PACS hardware, and which services, repairs and maintains it under an annual contract. "The great thing about a managed service," explains Henderson, "is that we don't have to care where the hardware or software comes from; it's up to the provider to ensure everything works together and to come in if anything breaks down. Everything is bundled in, so essentially all we're doing is babysitting the system."
Henderson sees the issue of integration as being the main driver for future developments, with PACS networks being extended to GP surgeries, for example, and voice-recognition systems and 3D recognition being integrated into them as well. "In the long-term, I can envisage more health trusts using PACS for all their imaging as well as archiving - it will become less a case of PACS storage, more a case of just storage, integrated for everything," he says.
And these systems look as though they're here to stay, because although some newer modalities such as MRI are making inroads into traditional techniques, few people believe we will see a day when a single modality will be used for all medical imaging.
As Srinivasan says: "In the future, I can see MRI systems being used more and more in different applications, but I don't see other modalities trailing away as a result."
Size and price restrictions
"What's holding them back at the moment are their physical size and price," Srinivasan continues. "Another key factor is the volume of clinical confirmations for different indications. For example, while there are MRI-based products to diagnose breast cancer, it is still not the predominant diagnostic tool. As the case histories build up, however, that situation is likely to change."
This is echoed by Darshana De, research analyst at business research and consulting firm Frost & Sullivan. "MRI is one of the most powerful and widely used diagnostic tools these days," she says. "At present, muskuloskeletal and abdomen imaging are the most popular applications, although the proportion of procedural volumes for cardiac and breast MRI is increasing - in fact, the market for breast MRI is set to boom over other applications."
There are a number of reasons for this wide and growing take-up of MRI. It is perceived as inherently safe - because it relies on electromagnetic radiation rather than X-rays, for example - and delivers high-quality image resolution and contrast. Another key advantage is that it gives anatomical as well as functional imaging - blood flow, cardiac function and so on - whereas the other modalities tend to have strengths in only one or the other. X-rays, for example, show only the anatomy, while PET gives only functional information.
MRI doesn't have it all its own way though. The systems and set-ups they need are extremely expensive - of the order of hundreds of thousands of pounds - so only the biggest hospitals can justify having one. And, as anyone who has had an MRI scan will vouch, they can be a pretty claustrophobic experience.
But while the high cost will remain an inhibiting factor for some time to come, progress is being made in replacing conventional, 'closed' MRI systems with Open MRI, which many physicians prefer because they help to eliminate patients' concerns over claustrophobia and help to circumvent the growing problem of overweight patients.
An open MRI system is typically based on a C-arm or four-point scanner, using magnets at the end of each point situated around the patient. But while the sides are open, says Frost & Sullivan's De, the positioning of the magnets at the sides of the patient can make an open system as confining as a closed one. Also, the strength of the magnetic field in an MRI scanner determines the speed of the scan and the quality of the image, and while closed MRI systems use fields up to 3 Tesla, commercially available open MRI machines tend to have field strengths of between about 0.2 and 0.6 T. This leads to a lower signal-to-noise ratio and therefore poorer image resolution, so there is clearly still work to be done on bringing open MRI on a par with closed systems.
Another area of research into MRI is real-time control, and one researcher focusing on this is Dr David Larkman, of Imperial College, London.
His personal interest from an engineering standpoint is in removing the boundary between the scanner and the patient. He explains: "One of the big changes in MRI will be the potential for real-time control. If we can use the real-time aspects of control to get feedback from the patient on, say, their heartbeat and lung function then we can inform the scanner on how to operate scans to produce images free from artifacts."
Current MRI systems typically take several minutes to perform a scan, so the patient's functions cause what's called 'motional blurring' of the image - rather like the blurring of running water in a photograph taken using a slow shutter speed. "So we're developing a technology called parallel imaging, equipping an MRI scanner with multiple receivers and transmitters to give, say, 16 or 32 channels of information," says Dr Larkman. "These multiple inputs and outputs give us a lot of flexibility and control, so the challenge at the moment is to marry them up to physiological information.
"But the algorithms to do this are quite intensive and are not yet really up to running in real-time, despite the computing horsepower that's out there. So the key issue is getting the algorithms to run fast enough for real-time control."
Dr Larkman thinks we will see this technology in use in five to seven years' time. And in roughly the same time frame, medical imaging looks set to benefit from at least one advance in automation. Srinivasan at Cambridge Consultants says: "Looking ahead ten years or so, one development I can see making inroads is computer-aided diagnosis of medical images, as it promises to take some of the labour out of image analysis."
The technology already exists, and an obvious application for it is in mass breast screening programmes, but at the current stage of its development it has its detractors. Henderson at UHW says: "It's too sensitive at the moment. It has to be double-reported, by the computer and a qualified person. And I have reservations about its take-up in the UK because of the strong position radiologists hold."
But Srinivasan counters this: "Taking some of the labour out of image analysis does not diminish in any way the crucial role radiologists play. As an engineer, my hope and goal would be to put a powerful tool in the hands of the radiologist, and I believe the data volumes and case loads would lead radiologists to welcome such aids."