Resolution Recovery in PET During AWOSEM Reconstruction: A Performance Evaluation Study

• Published in: IEEE transactions on nuclear science Vol. 54; no. 5; pp. 1626 - 1638
• Main Authors: De Bernardi, Elisabetta, Mazzoli, Marco, Zito, Felicia, Baselli, Giuseppe
• Format: Journal Article
• Language: English
• Published: New York IEEE 01-10-2007; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Delay; Image contrast; Image reconstructiom; Image reconstruction; Image resolution; Imaging phantoms; Iterative algorithms; Mathematical models; Monitoring; Neoplasms; Noise; Oncology; ordered subset expectation maximization (OSEM); Positron emission tomography; positron emission tomography (PET); Reconstruction; Recovery; Spatial resolution; Tuning
• ISSN: 0018-9499; 1558-1578
• DOI: 10.1109/TNS.2007.905167

Lesion detectability and quantification in fluorine-18-fluorodeoxyglucose positron emission tomography oncological studies are highly affected by space variant blur. The performance of spatial resolution recovery during AWOSEM iterative reconstruction was assessed by including in the system matrix a simple model of point spread function fitted on experimentally acquired point source sinograms. The model described axial, tangential and radial blur; radial asymmetry was taken into account. The algorithm properties in terms of contrast recovery, image noise and object separability were evaluated on scanned sphere phantom studies both in 2-D and 3-D mode. Extensive comparisons of standard and resolution recovery algorithms, over wide parameter ranges with and without post-filtering are presented. Resolution recovery, by delaying noise breakup appearance and by increasing contrast, was able to improve overall accuracy, particularly for small objects and large blur. The recovery of axial blur resulted determinant in 3-D and even more in 2-D mode. The optimal choice of reconstruction parameters was shown to be highly object dependent, thus suggesting an algorithm tuning according to applications.