An efficient voxel-driven system model for helical pinhole SPECT

Conventionally, a pinhole SPECT acquires data by rotating the Gamma-camera in a circular orbit. However, such a acquisition method could merely scan a very limited field-of-view (FOV). For a longer object, it requires several circular acquisitions over various bed positions to cover the entire object. Multiple-bed acquisition also requires a high accurate movement control to avoid the overlapping or gap between any adjacent bed positions. In this work, we propose a voxel-driven (VD) system model (projection matrix) for iterative helical pinhole SPECT image reconstruction. The photon detection along helical trajectory can reduce acquisition time without multiple-bed overlapping and gap problems. However, the most challenging part for helical pinhole SPECT is the increase in image volume. For efficient computation and matrix storage reduction, the geometric symmetries derived from helical trajectory shall be fully utilized. By using a 16-fold symmetry derived from helical trajectory, the system matrix requires only 90 degrees information of projection geometry. A multi-threaded code can be implemented to accelerate the projection computation. The result for a two line sources phantom study shows that the VD model is efficient and suitable for iterative helical pinhole SPECT. The VD matrix could model the helical trajectory pinhole SPECT efficiently and handle a longer object without increasing any storage space of system matrix. The proposed model is also ready to be extended multiple-pinhole system for to better sensitivity as well as other fast reconstruction algorithms.