Motion compensated reconstruction of 4D SPECT using parallel computation and deformable content adaptive mesh

In this abstract, we present preliminary evaluation of a full 4D reconstruction method aimed at cardiac gated emission tomography. The proposed method uses content adaptive mesh model (CAMM) to represent the 3D volumetric data. The nodes of the CAMM are then moved from a gate to gate, according to the motion obtained from a motion estimation algorithm, to form 4D deformable content adaptive mesh model (DCAMM). The use of a CAMM as well as DCAMM allows incorporating all data degradation models, namely objecting attenuation and detector-collimator spatial response, referred to as distance dependent blur. The parallel implementation, using OpenCL, was essential to shorten the needed computational time and make this a practical reconstruction methodology. In this work we successfully tested proposed parallel implementation by reconstructing images obtained from a realistic cardiac gated SPECT simulation using SIMIND Monte Carlo package and NCAT phantom.