ITBusiness.ca

Non-profit organization uses SGI supercomputer to model human organs

Until now, neurosurgery has been the traditional application for image-guided surgery.

But Robarts Research Institute and medical visualization software developer Atamai Inc., both of London, Ont., along with Mountain View, Calif.-based Silicon Graphics Inc., are working on a two-year project

at the institute’s Virtual Augmentation and Simulation for Surgery & Therapy (VASST) lab to make image-guided surgery a reality for other types of medical procedures, such as prostate cancer therapy, breast cancer biopsy and cardiac intervention and therapy as well.

The institute is using a 20-processor Altix 3000 server, a dual-processor Altix 350 server with 20 GB of memory, and a 6 TB InfiniteStorage Storage Area Network (SAN) solution from SGI to model functioning human organs to aid in the development of minimally invasive surgery and therapy techniques.

“”Neurosurgery was always the easiest place to start because your skull forms a nice holder for your brain and it’s easier to create a registration, which basically is what allows the computer to point to something you can’t see and show you where to make to make the holes so you end up taking the shortest path to the target,”” explains Atamai president Yves Starreveld. Atamai, he says, initially started as a joint venture between Robarts, the University of Western Ontario, London Health Sciences, Starreveld, David Gobbi and another PhD student, Kirk Finnis.

“”If you can imagine general surgery it’s like you have a bunch of guts in a cavity and of course they are going to be moving around,”” adds Atamai lead software developer Gobbi. “”If you do surgery on a heart, the heart is continuously beating, so simulating general surgery and cardiac surgery is a much more challenging problem than doing the same for brain surgery.””

Dr. Terry Peters, scientist and principal investigator at Robarts, a non-profit research organization, says the project grew out of his work at the Montreal Neurological Institute and Hospital on minimally invasive image-guided surgery on the brain.

“”I was approached by cardiac surgeons who had begun to perform coronary artery bypass procedures using robots and they felt they needed sophisticated imaging infrastructure to help them plan and guide these procedures,”” says Peters. “”Based on the work we had been doing in those areas, along with my colleagues at Robarts, we decided we should move into the area of modelling organs so we could set up an environment that would allow us to develop new, minimally invasive procedures using organ models within the computer, rather than having to do an excessive amount of experimentation on animals.””

The institute was awarded a $2 million grant from the Canadian Foundation for Innovation, another $2 million from the Ontario Innovation Trust and another $1 million from private sources and other smaller grants.

Robarts opted for SGI technology over that of other supercomputer providers due to the vendor’s long experience in medical visualization, he says.

“”In many instances SGI has more or less set the industry standards for medical visualization, so we had confidence the infrastructure at SGI was there to provide solutions to our problems,”” he says. “”Second, it was attractive to us to adopt a Linux-based supercomputer that uses the same operating system that most of the students use in the lab to develop software on the desktop, so we now have a seamless research environment which goes from single processor systems on various desktops to a multiprocessor supercomputer, all using the same operating system.””

According to Starreveld, Atamai’s task was to develop software that would allow the VASST lab to harness the power of the SGI supercomputer to enable researchers to apply it to the task at hand. “”They didn’t just want something that would do the one task really nicely, they actually wanted bits and pieces of software that would allow other researchers and grad students to develop those applications, so our job is really to provide a layer on top of the computer and to allow people who might not have sophisticated computer knowledge or abilities to develop these apps without having to worry about the intricacies of the machine itself,”” says Starreveld.

SGI’s task, says Colin Holmes, business development manager for SGI’s science and technology industry, was to not only provide the hardware but to write the code that makes the best use of that hardware. Holmes says while the challenge 10 to 15 years ago was to compute images based on scanner technology, today’s challenge is to manage all the images that are generated in a health care setting, espcially now that those images are being combined with each other.

“”Bringing all those sets of information together is a challenge — it’s data-intensive and it’s computationally intensive, because for each algorithm, every time someone invents a new way to look at a particular aspect of it you may need to rerun that algorithm for each of the subjects you’ve acquired, and these are often in the thousands of subjects,”” he says.

Over the last five years people have been very excited about the ability to answer yesterday’s questions using PC technology — but they’re still yesterday’s questions, he adds.

“”What SGI likes to do is say, ‘what if you had a machine that was10 times more powerful than that? What questions would you ask it then?’ And that leads to some of the most exciting discussions we have with these scientists.””

Exit mobile version