Photon-counting detectors can generate energy-specific images that are assigned to energy bins in relatively small ranges. The trade-off is that because each bin uses only a small portion of the available photons, image noise is greater.
This led Zhicong Yu, PhD, and colleagues to develop an iterative reconstruction algorithm aimed at reducing the noise back to normal levels.
"The purpose of this work was to develop and validate a compressed sensing iterative reconstruction technique that could reduce the noise in the bin images to levels close to that of the composite image (which uses all detected photons)," wrote study co-author Cynthia McCollough, PhD, in an email to AuntMinnie.com.
Using their spectral prior image constrained compressed sensing (PICCS) algorithm, the researchers obtained significant noise reduction, averaging 70% in the energy bin images, to the benefit of spectral CT imaging applications.
Check out this Monday afternoon session to get all the details.
