Yannan Jin MSC, Presenter: Nothing to Disclose

Yiannis Kyriakou PhD, Abstract Co-Author: Nothing to Disclose

Willi A. Kalender PhD, Abstract Co-Author: Consultant, Siemens AG Consultant, Bayer AG Founder, CT Imaging GmbH Scientific Advisor, CT Imaging GmbH Shareholder, CT Imaging GmbH Founder, Artemis Imaging GmbH CEO, Artemis Imaging GmbH Shareholder, Artemis Imaging GmbH

The aims of this study are (1) to suppress the noise in energy-resolved CT using highly-constrained back-projection (HYPR) algorithm and (2) to assess the effect of HYPR on the image quality parameters of decomposed images such as iodine or virtual non-contrast (VNC) images.

We focused on the simulation of an experimental scanner equipped with a photon counting detector with eight energy bins. The third and forth energy bin were separated at 33 keV, the k-edge of iodine. The energy-resolved HYPR algorithm was implemented in two steps: 1) calculate the low-noise composite image using the photons in all energy bins; 2) calculate the energy weighting function with the constraint of the composite image. The spatial resolution is preserved as the energy weighting function is only weakly correlated to the spatial components. After the noise suppression, a linear regression method was applied for material decomposition. A digital mouse phantom with bone, iodine and soft tissue inserts was used to validate the proposed algorithm. For comparison, we also measured the physical mouse phantom on a dual-source dual energy micro-CT scanner (TomoScope Synergy Twin, CT Imaging GmbH, Erlangen, Germany) and implemented an image-based three material decomposition method. The tube was operated at 40 / 65 kV with 0.5 mm aluminum filter and the tube current was adjusted so that the total dose was equivalent for both scans.

The simulation results show that on average, the noise level was reduced by 40%-50% in each energy bin when HYPR was applied. The contrast of inserted materials was well preserved. For decomposed images, the proposed algorithm reduced the noise level of the VNC image by about 40% while maintaining the error of decomposition below 3%. Compared to the dual-source dual energy micro-CT, the noise level of iodine and VNC image acquired from the energy-resolved CT with HYPR was lower by 80% and 60%, respectively.

The HYPR algorithm substantially reduced the noise level of both reconstructed and decomposed images while maintaining the contrast and decomposition accuracy in energy-resolved imaging.

The HYPR algorithm can improve the image quality of both original and decomposed images in energy-resolved CT or reduce the dose at the same noise level.

Jin, Y, Kyriakou, Y, Kalender, W, Noise Suppression in Energy-resolved Computed Tomography Using Highly-constrained Back Projection (HYPR).  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL. http://archive.rsna.org/2010/9006987.html