Background correction strategies for list-mode submillimetre 3D image reconstruction algorithm applied to the high resolution Quad-HIDAC PET system

Two novel methods for the inclusion of the background estimation within the list-mode reconstruction process are presented. The additive and subtractive schemes for background correction are very well known procedures included within the reconstruction model. The first strategy proposed includes the background estimation using an additive model, in this case the background counts resulted from the forward step over the background image, previously reconstructed, is added to the forward step over the estimated image. The second strategy proposed, changes the model “1/q”, typically used in the event by event reconstruction process, to “(1-(b/m))/q”, in which the numerator is the event probability of being a true or a background event. A
¹⁸
F small mouse NEMA phantom is used to assess the methods along with a clinical list mode data from a small mouse scan. Both sources were acquired from the stationary quad-HIDAC PET scanner. Quality image comparisons and noise investigation between the two models are presented to demonstrate their performance.