Study on Iterative Reconstruction Methods for Animal Ct

X-ray computed tomography (CT) offers anatomical images of attenuation characteristic within living bodies, and if combined with nuclear medicine imaging, provides both functional and anatomical images for accurate diagnosis. CT images can be also used to correct the attenuation in the nuclear medicine images with better quality of attenuation coefficients. The new developed multi-slice CT system with helical scanning has the advantage of fast scanning and low dose, thus improving spatial and temporal resolutions, and also offers better dynamic information. An image reconstruction method is needed to convert the scanned projection data into a tomographic image in CT. Conventional analytical reconstruction method, FBP (Filtered Backprojection), is used widely due to its fast reconstruction speed. However, under the conditions of low energy, low dosage, and fast scanning, the image quality from FBP reconstruction in modern CT is sub-optimal due to many reasons. Iterative reconstruction method can overcome the above problems with correct noise and physical modeling, and can be readily adapted to different geometrical systems. It has been proved that iterative reconstruction outperforms FBP with better image qualities. We plan to build up an iterative reconstruction platform for more qualitative CT reconstruction for the animal CT system developed in INER. We will study different noise treatment and then build up a system model for cone-beam CT and an iterative CT reconstruction using an OS-type convergent algorithm. This system will facilitate the physical effect corrections including scatter, metal artifact and beam-hardening, in the future. In particular, we will use the reconstruction platform in this proposal as a development base for the future helical CT system in INER.

Project ID:PG9504-1276
External Project ID:952001INER055