Objectives: A novel approach to the PET imagereconstruction is presented, based on the inclusionof image deconvolution during conventional OSEMreconstruction. Deconvolution is here used to providea recovered PET image to be included as a prioriinformation to guide OSEM toward an improvedsolution.Methods: Deconvolution was implemented usingthe Lucy-Richardson (LR) algorithm: Two differentdeconvolution schemes were tested, modifying theconventional OSEM iterative formulation: 1) We builta regularizing penalty function on the recovered PETimage obtained by deconvolution and included itin the OSEM iteration. 2) After each conventionalglobal OSEM iteration, we deconvolved the resulting PET image and used this recovered version as theinitialization image for the next OSEM iteration.Tests were performed on both simulated andacquired data. Results: Compared to the conventional OSEM, boththese strategies, applied to simulated and acquireddata, showed an improvement in image spatialresolution with better behavior in the second case.In this way, small lesions, present on data, could bebetter discriminated in terms of contrast. Conclusions: Application of this approach to bothsimulated and acquired data suggests its efficacyin obtaining PET images of enhanced quality.
Using Deconvolution to Improve PET SpatialResolution in OSEM Iterative Reconstruction
Rizzo G;Castiglioni I;Gilardi MC;Fazio F;
2007
Abstract
Objectives: A novel approach to the PET imagereconstruction is presented, based on the inclusionof image deconvolution during conventional OSEMreconstruction. Deconvolution is here used to providea recovered PET image to be included as a prioriinformation to guide OSEM toward an improvedsolution.Methods: Deconvolution was implemented usingthe Lucy-Richardson (LR) algorithm: Two differentdeconvolution schemes were tested, modifying theconventional OSEM iterative formulation: 1) We builta regularizing penalty function on the recovered PETimage obtained by deconvolution and included itin the OSEM iteration. 2) After each conventionalglobal OSEM iteration, we deconvolved the resulting PET image and used this recovered version as theinitialization image for the next OSEM iteration.Tests were performed on both simulated andacquired data. Results: Compared to the conventional OSEM, boththese strategies, applied to simulated and acquireddata, showed an improvement in image spatialresolution with better behavior in the second case.In this way, small lesions, present on data, could bebetter discriminated in terms of contrast. Conclusions: Application of this approach to bothsimulated and acquired data suggests its efficacyin obtaining PET images of enhanced quality.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
