Premium
Poster – 02: Positron Emission Tomography (PET) Imaging Reconstruction using higher order Scattered Photon Coincidences
Author(s) -
Sun Hongwei,
Pistorius Stephen
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4961776
Subject(s) - physics , photon , compton scattering , attenuation , iterative reconstruction , energy (signal processing) , optics , annihilation , detector , image resolution , correction for attenuation , range (aeronautics) , scattering , photon energy , reconstruction algorithm , computational physics , nuclear physics , computer science , computer vision , materials science , quantum mechanics , composite material
PET images are affected by the presence of scattered photons. Incorrect scatter‐correction may cause artifacts, particularly in 3D PET systems. Current scatter reconstruction methods do not distinguish between single and higher order scattered photons. A dual‐scattered reconstruction method (GDS‐MLEM) that is independent of the number of Compton scattering interactions and less sensitive to the need for high energy resolution detectors, is proposed. To avoid overcorrecting for scattered coincidences, the attenuation coefficient was calculated by integrating the differential Klein‐Nishina cross‐section over a restricted energy range, accounting only for scattered photons that were not detected. The optimum image can be selected by choosing an energy threshold which is the upper energy limit for the calculation of the cross‐section and the lower limit for scattered photons in the reconstruction. Data was simulated using the GATE platform. 500,000 multiple scattered photon coincidences with perfect energy resolution were reconstructed using various methods. The GDS‐MLEM algorithm had the highest confidence (98%) in locating the annihilation position and was capable of reconstructing the two largest hot regions. 100,000 photon coincidences, with a scatter fraction of 40%, were used to test the energy resolution dependence of different algorithms. With a 350–650 keV energy window and the restricted attenuation correction model, the GDS‐MLEM algorithm was able to improve contrast recovery and reduce the noise by 7.56%–13.24% and 12.4%–24.03%, respectively. This approach is less sensitive to the energy resolution and shows promise if detector energy resolutions of 12% can be achieved.