Cell Broadband Engine, the workhorse powering Playstation 3 consoles, is a powerful multi-core microprocessor specifically developed to operate in rich visual environments of video games. Medgadget has learned that IBM, the maker of the chip, alongside Mayo Clinic, has been working on implementing the chip in medical imaging software to speed up object recognition and increase image precision that is derived from multiple data sources.
From an upcoming press release, obtained by Medgadget:
Collaborators from Mayo Clinic and IBM have exploited parallel computer architecture and memory bandwidth to dramatically speed the processing of 3-D medical images. The advance significantly aids image registration — the computer-enhanced alignment of two medical images obtained at different dates or by using different imaging devices, in three-dimensional space. With the images properly aligned over one another, a radiologist can more easily detect structural changes such as the growth or shrinkage of tumors.
The results will be presented in full in a joint presentation by Mayo Clinic and IBM at the IEEE (Institute of Electrical and Electronics Engineers) International Symposium on Biomedical Imaging in Washington, D.C., April 12-15.
“This alignment of images both improves the accuracy of interpretation and improves radiologist efficiency, particularly for diseases like cancer,” says Mayo radiology researcher Bradley Erickson, M.D., Ph.D.
Through porting and optimization of Mayo Clinic’s Image Registration Application on the IBM® BladeCenter® QS20 ‘Cell Blade,” the application produced image results fifty times faster than the application running on a traditional processor configuration. [Cell Blade is pictured –ed.]
One way medical images are being improved is by using visual images from more than one source — magnetic resonance imaging (MRI) and computerized tomography (CT) scans for example. The generation of computer-enhanced images from multiple sources must begin with accurate alignment of the visual data. When three dimensions and millions of pixels are involved, the task becomes exponentially complex. Within this scope, the need for higher processing speeds is essential.
For this imaging project, Mayo Clinic and IBM used 98 sets of images and ran the optimized registration application on the IBM BladeCenter QS20, in comparison with running the original application on a typical processor configuration. The application running on a typical processor configuration completed the registration of all 98 sets of images in approximately 7 hours. The team adapted a “mutual-information-based” 3-D linear registration algorithm application optimized for Cell/B.E. and completed the registration for all 98 sets of images in just 516 seconds, with no registration taking more than 20 seconds.
The 3-D linear algorithm finds the best spatial positioning to maximize the amount of information gathered from the two images, thereby optimizing sampling quality while reducing sampling time. Greater efficiencies were achieved by caching data in cuboids or “bricks” so image sampling did not “waste” pixels. When sampling ratio was comparatively low, the team packed the sampled moving pixel images in a contiguous fashion (in an “image stripe”) to speed retrieval when needed.
Considering the continued proliferation of CT scanners, MRI machines, and other tomographic devices, this looks like a sound move for IBM to help to get more out of the data produced by these machines.
UPDATE: Press release: Mayo Clinic and IBM Score Significant Advance in Real-Time Medical Imaging …