Grid technologies for the medical community
Powerful computer simulation tools have been developed to assist doctors in diagnosis, pre-operation planning and surgery. So powerful in fact that many of these tools cannot be run efficiently on normal computers. The Grid, however, is much more than a normal desktop - it is a vast interconnected collection of computers, programmes and people. And the IST project GEMSS is harnessing the Grids processing power to place it in the hands of medical practitioners.
The GEMSS project plans to present the first prototype of its Grid middleware at the end of February along with a testbed that will be one of Europes first computing and resource Grids for clinical use, allowing easy access to advanced simulation and image processing tools operating at levels of speed and efficiency that conventional local hospital systems cannot match. Developed by 10 partners from academia and industry, GEMSS (Grid-Enabled Medical Simulation Services) incorporates tools designed by previous European medical projects such as BloodSim, SimBio, COPHIT and RAPT that created effective but complex and computationally demanding aids.
"Simulation and planning tools are difficult to handle and need large amounts of computing resources to be of use and provide output you can trust," explains project coordinator Jochen Fingberg of NEC Europe. "Simulation and image processing tools, such as those incorporated into GEMSS, have been largely underused until now because of that." Simply, they need more IT resources than most hospitals and clinics can afford or accommodate.
Although computer processing power and memory capacity have advanced rapidly in recent years - the raw speed of individual computers is now one million times faster than it was 50 years ago - cost-efficient local platforms are still far too slow and distributed systems are too complex to handle to solve scientific problems, not least in the medical sector where time and accuracy are crucial. The obvious solution is the Grid - a Web on steroids that harnesses the processing power of multiple computers to allow seamless access to computing resources and services over the Internet.
Testing a range of applications
The GEMSS testbed comprises six different medical applications, ranging from a tool to simulate inhaled drug delivery, which resulted from the COPHIT IST project, to a cardiovascular modelling system based on another IST project, BloodSim. By choosing a variety of different but equally highly demanding applications, the GEMSS consortium expects to prove the Grids potential to improve planning and diagnosis in many medical fields through predictive high-performance simulation and advanced image processing.
Following the presentation of the first prototype in February the partners will go on to carry out a series of pilot trials with medical practitioners aimed at validating the GEMSS testbed and middleware. Expectations are high. For example, the project estimates that a planning support system to model bone structures that assists in Maxillo-facial surgery could be run in as little as 10 minutes with the power of the Grid, compared to as much as four hours on an individual machine. The trials, involving neurosurgery, cardiovascular, cranial and respiratory system simulation tools, are expected to confirm such benefits and validate the medical relevance of the numerical models in the GEMSS applications.
"These time savings are crucial for the effective use of some tools," Fingberg notes. "For example the neurosurgery tool is designed to be used during surgery when the patient is on the operating table." Evidently, for both patient and surgeon, the faster data is processed the better.
But while seamless access to services and speed are important factors, so too is security.
Given the sensitive nature of the information, from bone scans to brain scans, the data required for the simulations must be processed in a secure and lawful way, a basic necessity if Grid technologies are to be employed by the medical sector.
"I dont think the Grid is being used practically in the medical community at present - the main reason being security and legal issues," Fingberg says, noting that in evaluating existing Grid systems the project found that generic middleware provides insufficient guarantees of security and privacy.
A secure and lawful approach
GEMSS was therefore designed with a service-oriented structure that makes security easier to address, and transport level as well as end-to-end security features have been incorporated into the middleware, which will be compliant with EU laws.
For the medical sector, secure Grid technologies promise to vastly improve procedures and patient care, and perhaps most significantly, if they are to be adopted commercially, they can also provide important financial savings.
To obtain the computational power needed to efficiently use applications such as those incorporated into the GEMSS testbed, clusters of computers and IT infrastructure including expertise would be needed - something only the largest research institutions could afford. With the Grid, however, all a hospital, clinic or individual practitioner needs is a desktop computer connected to a Grid service provider via the Internet. Medical practitioners would only pay for processing power as and when they require it, rather than investing in infrastructure that is used infrequently and doesnt allow the accommodation of new services at low cost.
But when will use of the Grid itself become commonplace in the medical sector?
"One has to be patient," Fingberg says. "We need more pilot studies that test more applications in more areas of healthcare to convince the medical community of the benefits. I think it will be at least three to five years before the medical sector makes real use of the Grid."
Set to conclude in February 2005, the GEMSS project represents an important step toward ensuring that happens.
Tara Morris | alfa