Stanford MDCT conference: Spectral CT technique mines signal for additional information

It's not what you get, it's what you keep could be an idiom that applies directly to imaging technology.

For years, ultrasound had been throwing away data, information buried in the returning echo signal, because physicists didn't know what to do with it. One day, the light bulb went on above someone's head, and slow-flow imaging was born.

The same process may now be happening in CT.

GE Healthcare will describe today at the International Symposium on Multidetector-Row CT early work to develop a new kind of CT detector that does more than count individual photons. This detector, already in the prototype stage and generating phantom images, translates photon strikes into anatomic position and material characteristics. Among the possible distinctions it will try to make are contrast media, calcifications, and bone versus soft tissue, particularly blood vessel walls.

"Ultimately, we want to use the same x-ray beam, discerning from it what the (body) materials are, in a way that doesn't overexpose patients to dose," said Gene Saragnese, global vice president of molecular imaging and CT at GE Healthcare. "So we are looking at a technique by which we count the photons and characterize their energies as they come in."

The technique, called spectral CT, focuses on getting as much information from each photon striking the detector as possible, according to Saragnese. Because the signal contains so much information, this particular technique prefers low dose.

"It doesn't work very well if you slam the detector with the kind of flux levels seen in today's CT scanners," said Brian J. Duchinsky, global general manger for GE Healthcare's CT business.

Best of all for the engineers and bean counters at GE, the detector that makes this happen requires standard materials, even though the data harvest is much greater.

"We are the only company thinking that way, at least up to this point," Duchinsky said.