A Direct Ray Tracing Reconstruction Algorithm Using an Adaptive Median Filter

A direct ray tracing algorithm has been proposed under the development of a dedicated brain PET insert scanner, called MindView. The presented method takes advantage of the quasi- continuum nature of the impact coordinates provided by detectors composed by of monolithic scintillation crystals (LYSO material). This method requires a two-step filtering process: first, a Butterworth filter, multiplied by a ramp filter, is applied on the reconstructed image to wipe out the low oversampled frequencies after performing a backprojection, and second, an adaptive median filter is applied to the image for the noise to be reduced. Our algorithm, has been evaluated with simulation and real data, and compared with some of the gold-standard algorithms, in terms of image spatial resolution, such as the Filtered Backprojection 3D Reprojected (FBP3DRP) or the List Mode Ordered Subset (LMOS). Data obtained with a Derenzo-like phantom was used showing the 1.6 mm rods and a mean peak-to-Svalley ratio of 2.4 with the proposed algorithm, while in the FBP and LMOS depicted 1.3 and 1.4, respectively.