Mercury Computer Systems leverages NVIDIA Technology alliance to accelerate DBT reconstruction
MARCH 7--Mercury Computer Systems Inc. (Chelmsford, MA; www.mc.com) and NVIDIA Corporation (Santa Clara, CA; www.nvidia.com) announced during the European Congress of Radiology in Vienna, Austria, the first results of Mercury's partnership with Massachusetts General Hospital (MGH), in which Mercury greatly reduced the image-reconstruction time for the 3-D mammography technique developed by MGH from five hours to five minutes--a 60X performance increase using NVIDIA technology.
Traditional mammography imaging relies on 2-D x-ray images in which growths can be obscured and undetected. Digital breast tomosynthesis (DBT) reconstructs a 3-D volume from a series of 2-D projection images taken over an arc of 50°. With DBT, physicians can "page through" the interior of the breast without the superimposition of the other tissues.
The DBT method involves vast amounts of computation, which previously took far too long to be clinically viable. Mercury engineers worked with the breast imaging division at MGH to enhance its innovative technique using Mercury image-processing expertise and mathematical optimizations, greatly reducing the image-processing time and preserving high image clarity. Combined with a unique image-processing solution that uses enhanced NVIDIA Quadro professional graphics processor technology, Mercury achieved unprecedented image reconstruction results.
"DBT will enable physicians to find more cancers earlier, while simultaneously reducing many of the false alarms that occur using conventional mammography," said Dr. Daniel B. Kopans, director of the breast imaging division at MGH. "This will reduce the overall cost of breast-cancer screening while improving its accuracy. Moreover, the DBT technique will eliminate the need for patients to endure additional examination time or x-ray doses."
Mercury and MGH are working to move the DBT imaging technique from the laboratory to widespread clinical use and are continuing their collaboration to improve cancer detection and diagnosis.
