Conventional reconstructive algorithms assume that the x-ray source and detector are points and that the x-ray beam is a line. Last year, GE Healthcare introduced scanners that use an algorithm called adaptive statistical iterative reconstruction (ASIR). The technique compares neighboring voxels; if one looks too different, it is assumed to be noise and is removed. So the scanner can use less intense x-rays, which can result in more noisy images. The technique can cut the radiation dose of a CT scan by half.
The new algorithm by Yadava and his colleagues goes one step further. It uses a more realistic physics model of the x-ray source, the detectors, and the x-ray beam. Each of these three is assumed to have specific diameters instead of being considered a point or a line, Yadava says. Depending on the type of scan, the technique is better than ASIR at cutting image noise, and thus the x-rays can be even less intense. The researchers got high-quality abdomen scans of a human model using an eighth of the radiation dose of a conventional scan.
Software techniques are one promising way to reduce radiation, McCollough says. Over the past decade, other techniques have reduced the radiation dose from CT scans–abdomen scans now deliver a third of the dose that they delivered in the 1980s. One major advance was adding multiple detectors to the scanners, she says. The other was to adjust the x-ray intensity depending on the patient’s size and the organ being imaged. This has cut radiation doses by up to 40 percent.
McCollough has developed another dose-reduction technique that involves automatically adjusting the energy spectrum of the x-rays. She says the method could give better quality images for children and thinner adults and use 20 to 40 percent less radiation. Many hospitals do x-ray energy adjustment manually right now, she says, but a major scanner manufacturer is incorporating the automatic technique into its machines.