Fast iterative reconstruction for helical pinhole SPECT imaging
Pinhole SPECT for small animal has become a routine procedure in many applications of molecular biology and pharmaceutical development. There is an increasing demand in the whole body imaging of lab animals. A simple and direct solution is to scan the object along a helical trajectory, similar to a helical CT scan. The corresponding acquisition time can be greatly reduced, while the over-lapping and gap between consecutive bed positions can be avoided. However, helical pinhole SPECT inevitably leads to the tremendous increase in computational complexity when the iterative reconstruction algorithms are applied. We suggest a novel voxel-driven (VD) system model which can be integrated with geometric symmetries from helical trajectory for fast iterative image reconstruction. Such a model construction can also achieve faster calculation and lower storage requirement of the system matrix. Due to the independence among various symmetries, it permits parallel coding to further boost computation efficiency of forward/backward projection. From phantom study, the results also indicate that the proposed VD model can adequately model the helical pinhole SPECT scanner with manageable storage size of system matrix and clinically acceptable computation loading of reconstruction.
