CSL Main >> Research >> Body Scans: More Speed, Less Radiation

Body Scans: More Speed, Less Radiation

New Algorithms Speed Image Reconstruction

By Doug Peterson

May 2005

CT scanning, the system widely used for full-body scans at hospitals, will get a big boost in both accuracy and speed, thanks to mathematical methods developed by CSL researchers.

As a bonus, their new methods could even reduce radiation exposure for patients, significantly cutting long-term health risks.

Yoram Bresler, a CSL professor, and Jeffrey Brokish, a research scientist, have worked with colleagues to develop algorithms that can provide nearly instantaneous, real-time reconstruction of complex, three-dimensional images -- procedures that normally would take several minutes.

Computer tomography (CT) scanning, which is still popularly known as CAT scanning, was first developed in the 1970s. At that time, Bresler said, it would take weeks to reconstruct a 3-D image out of many 1-D image “slices.”

With the newer machines, patients can be scanned from head to toe in 10 seconds, generating 4,000 image slices. But even though information collection has been sped up tremendously, reconstructing those images still lags behind, taking several minutes.

“Our goal has been to speed the image reconstruction so it happens at the same time the images are created,” Bresler said.

Real-time image reconstruction is vital in such medical procedures as fluoroscopy, in which doctors push a catheter through the heart.

“Doctors need to be able to see the catheter as it’s moving,” Bresler explained. “Doctors cannot tolerate a delay in image reconstruction or they will not know where the catheter exactly is at any given moment. Currently, they are only able to get 2-D or lower-resolution 3-D images in real time. With our method, it should be possible to get full 3-D resolution in real-time.”

Increased speed will also play a role in Homeland Security because it will help security agents scan more packages. Right now, Bresler said, a lot of packages coming into the United States on trucks are not scanned because the time it would take is impractical.

“With our algorithms, we can speed up image reconstruction by a factor of 10 to 50,” he added.

Not only would the image reconstruction be done faster, but resolution would be much better. Perhaps even more significant, however, patients may receive less exposure to x-rays in the long run.

People are exposed to radiation from many sources throughout their life. But according to Bresler, CT scans alone account for 60 percent of the radiation dose received by patients over their lifetime.

“This amount potentially accounts for 5,000 more deaths in the United States per year due to radiation exposure,” he said.

Bresler said the radiation dose can be reduced if the machines are able to reconstruct the data many times over, correcting the result each time. With the speed boost due to their algorithms, CT scanning systems will be able to do just that; they will be able to compute information many times over at blazing speeds, reducing the required radiation dose.

Now that the research groundwork has been laid, Bresler and former CSL colleague David Munson, with Brokish’s participation, have begun a start-up company known as InstaRecon, short for “Instant Reconstruction.”

“We’re getting a good response,” Bresler said. “But it’s up to us to prove the commercial viability of this.”

Companies will be particularly interested in finding out whether using the new algorithms is cheaper than throwing brute force at the problem of speed.

“The conventional method is to put more computers -- more hardware -- on the problem to get more speed,” Bresler said. “But our methods speed things up algorithmically. We don’t use more hardware, just more clever mathematics.”

CSL News

home