Multicore programming aids breast cancer detection
25 September 2008Tweet
A company that provides a CAD package to analyse breast ultrasound images is improving its performance with the use of a multicore development platform that helps get the best from CPUs and GPUs.
Medipattern is using RapidMind in its B-CAD system to enhance the performance of its CPUs, while also exploiting the acceleration available from GPUs, with a resulting eight-times performance increase.
RapidMind, which produces the RapidMind Multicore Development Platform, has recently launched a set of RapidMind Imaging Extensions, aimed specifically at applications such as medical imaging.
B-CAD V1 is the first FDA certified computer-aided detection system for breast ultrasound imaging. B-CAD automatically analyses the physical characteristics of ultrasound images, helping physicians characterise each lesion. B-CAD helps improve accuracy, increases confidence, and reduces transcription costs for the physicians caring for the millions of women who undergo breast ultrasound procedures every year.
'Timely results are crucial to our customers, so Medipattern needed to achieve an order of magnitude in increased performance. While the latest multicore processors offer the potential to improve processing times, they are extremely difficult to program and often increase development costs and time to market,' said Jeff Collins, CEO of Medipattern Corporation. 'RapidMind’s platform, expertise and professional services, let us meet this objective.'