Statistical data acquisition model for GPU based MLEM joint estimation of tissue activity distribution and photon attenuation map from PET data

摘要

We provide improved maximum likelihood expectation maximization (MLEM) joint estimation of emission activity and photon attenuation from positron emission tomography (PET) data. Lines of response (LOR) are divided along their length into cells having equal length. MLEM computations assume all intersections between LOR cells and voxels have an intersection length of the LOR cell length. This way of discretizing the problem has the significant advantage of leading to MLEM update equations that have a closed form exact solution, which is important for fast, accurate and robust estimation.