Graphics Processors Help Lower CT Scan Radiation

A newly devised way to process X-ray data could lower by a factor of 10 or more the amount of radiation patients receive during cone beam computed tomography (CT) scans.

Cone beam CT (CBCT) plays a crucial role in image-guided radiation therapy (IGRT), a state-of-the-art cancer treatment. IGRT uses repeated scans during a course of radiation therapy to precisely target tumors and minimize radiation damage in surrounding tissue. Though IGRT has improved outcomes, the large cumulative radiation dose from the repeated scans has raised concerns among physicians and patients.

Reducing the total number of X-ray projections and the mAs (milliampere second) level per projection (by decreasing the X-ray generator pulse rate, pulse duration, and/or current) during a CT scan can help minimize patient's exposure to radiation, but the change results in noisy, mathematically incomplete data that takes hours to process using the current iterative reconstruction approaches. Because CBCT is chiefly used for treatment setup while patients are in the treatment position, fast reconstruction is a requirement, explained lead author Dr. Xun Jia, a University of California, San Diego (UCSD; USA) postdoctoral fellow.

The research was presented July 2010 at the 52nd annual meeting of the American Association of Physicists in Medicine (AAPM) in Philadelphia, PA, USA). Based on recent advances in the field of compressed sensing, Dr. Jia and his colleagues developed an innovative CT reconstruction algorithm for graphics processing unit (GPU) platforms. The GPU processes data in parallel--increasing computational efficiency and making it possible to reconstruct a cone beam CT scan in about two minutes. Modern GPU cards were originally designed to power three-dimensional [3D] PC graphics.

With only 20 to 40 total number of X-ray projections and 0.1 mAs per projection, the team achieved images clear enough for image-guided radiation therapy. The reconstruction time ranged from 77 to 130 seconds on an nVIDIA Tesla C1060 GPU card, depending on the number of projections--an estimated 100 times faster than similar iterative reconstruction approaches, according to Dr. Jia.

Compared to the currently widely used scanning protocol of about 360 projections with 0.4 mAs per projection, according to Dr. Jia the new processing method resulted in 36 to 72 times less radiation exposure for patients. "With our technique, we can reconstruct cone beam CT images with only a few projections--40 in most cases--and lower mAs levels,” he said. "This considerably lowered the radiation dose.”

The reconstruction algorithm is part of the UCSD group's effort to develop a series of GPU-based low dose technologies for CT scans. "In my mind, the most interesting and compelling possibilities of this technique are beyond cancer radiotherapy,” said Dr. Steve Jiang, senior author of the study and a UCSD associate professor of radiation oncology. "CT dose has become a major concern of medical community. For each year's use of today's scanning technology, the resulting cancers could cause about 14,500 deaths. Our work, when extended from cancer radiotherapy to general diagnostic imaging, may provide a unique solution to solve this problem by reducing the CT dose per scan by a factor of 10 or more.”

Related Links:
University of California, San Diego