The buzz on iterative reconstruction
Greetings from sunny San Francisco, and thanks for checking out the inaugural run of STAT READ, the ICI's one and only meeting and tradeshow blog! Cat Vasko, editor of, here to keep you up to date on all things MDCT. This morning's general session on dosing and radiation exposure included an in-depth look at the latest buzzword in cutting-edge CT technology: iterative reconstruction. Norbert Pelc, ScD, kicked off the discussion with a rundown of how IR works and what makes it different from filtered back projection, the algorithm type currently employed in reconstruction of CT images. As Pelc explains it, filtered back projection (henceforth known as FBP) is the fast-and-dirty way to reconstruct images -- it works, and it works quickly, but it relies on one very dangerous assumption: that the scanner's data measurements are perfect. "It’s very efficient, and the thing to keep in mind is that if the data are perfect, the image will be correct," Pelc said. But how often is the data truly perfect? Not very often. And that's where IR comes in. IR uses the data to guess at the image that created it, testing and perfecting its guess with each new (get ready for it) iteration. And according to Pelc, it's a technique whose time has come -- not least because IR creates fewer artifacts and clearer images, making it potentially highly useful in low-dose imaging or imaging of bariatric patients. But don't take his word for it. No sooner had Pelc taken a seat in the Grand Ballroom at the Hyatt Regency than Dennis Foley, MD, stepped up to take his place. Foley's been working with IR for some time now, and decided it was time to test the diagnostic potential of IR images against their FBP-produced counterparts. In a study with two readers, IR was consistently rated higher than FBP in image quality, spatial resolution, noise suppression and high-contrast and low-contrast detail. "Overall, the preference was given to IR," Foley said. Stay tuned -- iterative reconstruction, coming soon to a scanner near you! --Cat