TABLE OF CONTENTS The aim of PET image reconstruction is to achieve an image of the activity distribution within the patient from the recorded sets of acquired data. In a tomographic study, data are acquired by performing several sets of linear scannings of the examined slice at different angles: each set gives an activity profile. The acquired profiles are then used to generate an appropriate number of independent equations to mathematically solve, where the unknowns are the activity elements of a grid (the image matrix).
In 2D, the image matrix elements are called pixels, and each slice is a plane, without thickness. The number of detected events in a LOR is the sum of activity along the line of scanning in the plane ( line integral), and the activity is uniform in each pixel (the value associated to the pixel is the mean activity). In addition, the reconstruction program considers the resolution and sensitivity as independent from the depth of the annihilation event. The reconstruction should include a method for attenuation correction. In 3D-MS ( multi-slice), the image is obtained by simply overlapping 2D images of adjacent slices, therefore the assumptions used are the same. In 3D-PVI ( volume imaging), the slice thickness is considered and voxels ( volume elements) substitute pixels: the uniformity assumption is then true for the voxel.
A profile or a line integral does not contain the information on the depth of the annihilation events. An attempt to recover this information...
