Abstract
By using an external source, attenuation correction in positron emission tomography (PET) can be achieved by measuring a transmission scan, and then calculated the ratio of blank scan and transmission scan data as attenuation correction factors (ACFs). In order to increase throughput of scanners and to reduce motion artifacts, post-injection transmission scans are widely used in clinical routine. However, the additional radiotracer activities in the scanner’s field of view contributed from patient can lead to a biased calculation of ACFs. To improve the accuracy of ACFs, another feasible approach is to reconstruct transmission image and then calculate the ACFs by forward projecting the transmission image. In this research, a joint Poisson model of post-injection transmission scan is proposed to incorporate true transmission data as well as noises including random events and cross-contamination from the emission events of the patient. Based on the proposed model, the transmission image is then reconstructed using a pre-condition conjugate gradient algorithm under the paradigm of maximum a posteriori estimation. We use clinical data acquired from a CTI EXACT HR+ scanner for model validation. The experimental results have indicated that the voxel values of reconstructed transmission image are close to those ones of linear attenuation coefficients at 511KeV photon. The proposed image reconstruction method for post-injection transmission scan can result in generation of more accurate ACFs and lead to improvement of PET quantitation.