Faster medical image processing

IBM and the Mayo Clinic have successfully tested a 3D imaging application which is able to register medical images approximately 50 times faster than a solution running on a traditional processor configuration.

The solution uses IBM’s Cell Broadband Engine to allow radiologists to detect a change in the size of tumours in seconds rather than hours.

One way medical images are being improved is by using visual images from more than one source, such as magnetic resonance imaging (MRI) and computerised tomography (CT) scans. 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 and requires higher processing speeds.

The team adapted a ‘mutual-information-based’ 3D linear registration algorithm application optimised for the platform, which uses the same microprocessor as Sony's Playstation 3. The application running on a typical processor configuration completed the registration of 98 sets of images in approximately seven hours. The IBM and Mayo team adapted the application to the Cell processor and completed the registration for all 98 sets of images in just 516 seconds, with no registration taking more than 20 seconds.

The 3D linear algorithm finds the best spatial positioning to maximise the amount of information gathered from the two images, thereby optimising 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.

The team hopes that by running the application faster, a doctor will be able to make a quicker diagnosis and promptly begin appropriate treatments for patients.