3D PET image reconstruction based on Maximum Likelihood Estimation Method (MLEM) algorithm
Positron emission tomographs (PET) do not measure an image directly. Instead, they measure at the boundary of the field-of-view (FOV) of PET tomograph a sinogram that consists of measurements of the sums of all the counts along the lines connecting two detectors. As there is a multitude of detectors build-in typical PET tomograph structure, there are many possible detector pairs that pertain to the measurement. The problem is how to turn this measurement into an image (this is called imaging). Decisive improvement in PET image quality was reached with the introduction of iterative reconstruction techniques. This stage was reached already twenty years ago (with the advent of new powerful computing processors). However, three dimensional (3D) imaging remains still a challenge. The purpose of the image reconstruction algorithm is to process this imperfect count data for a large number (many millions) of lines-of-responce (LOR) and millions of detected photons to produce an image showing the distribution of the labeled molecules in space.