O33. The value of different reconstruction algorithms for quantification of FDG PET brain imaging. (September 2016)

• Record Type: Journal Article
• Authors: Moalosi, T.C.G.; Dupont, P.; Mix, M.; Warwick, J.; du Toit, M.; Doruyter, A.
• Is Part Of: Physica medica. Volume 32(2016)Supplement 2
• Page Start: 151
• Publication Date: 2016-09
• Subjects: Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2016.07.041
• Languages: English
• ISSNs: 1120-1797

Abstract

Abstract : Introduction: Reconstruction influences the quantitative results in PET imaging. The aim of this study was the evaluation of different image reconstruction parameters and their impact on quantification for 18 F-FDG PET of the brain. The reconstruction parameters studied were the number of iterations, smoothing levels (relaxation parameter lambda), and the use of time of flight (TOF) information. In addition, we investigated the effect of signal-to-noise ratio on each of these reconstructed brain images, as well as the effect of scan duration on image quality. Materials and Methods: A Philips® Gemini-TF Big bore PET/CT was used for acquiring data of a 3D Hoffman Brain phantom. Data was acquired for 25 minutes in list mode format after injection of 40 MBq FDG, and reconstructed with a voxel size of 2 × 2 × 2 mm 3 using two different iterative reconstruction algorithms: LOR-RAMLA and BLOB-OS. The number of iterations and subsets was varied successive from 3/33 (vendor default) to 30/33, acquisition scan duration from 1 to 25 minutes, lambda was selected as smooth (0.7) and normal (1.0), and TOF was switched on and off for BLOB-OS. The impact on image quality was analyzed in 15 cortical and subcortical brain regions (volumes of interest, VOIs) and for grey and white matter. Results: Contrast increased for all regions of the brain and for grey matter/white matter (GM/WM) ratio if the number of iterations increased. Image convergence was reached after fifteen iterations for all different algorithms. When varying the smoothing filter it was found that lambda 1.0 resulted in a faster convergence than 0.7. The coefficient of variation (COV) for all VOIs showed BLOB-OS with TOF to be superior to the other algorithms. The COV results for different scan durations showed only a minimal improvement after 5 minutes in high-activity regions (GM), and after 10 minutes in low-activity regions (WM). Conclusion: Based on phantom data 18 F-FDG brain imaging for 10 minutes and reconstructed with the BLOB-OS algorithm including TOF information with 15 iterations is optimal on the Philips Gemini TF Big bore PET/CT. Further analyses is planned using patient data to verify if these findings remain valid in a clinical setting.