Optimizing Pet-Ct Image Reconstruction Parameters for Diagnostic Imaging Using Non-Tof Algorithms



Jaber ZS¹ ; Fallahi B² ; Geramifa P² ; Ebrahimi M² Show Affiliations
Source: Biochemical and Cellular Archives Published:2018

Abstract

18F-fluorodeoxyglucose positron-emission tomography/computed tomography (FDG-PET/CT) reconstruction algorithms can have substantial influence on quantitative accuracy especially for diagnostic oncology. The goal of this study was to optimize PET-CT image reconstruction parameters for diagnostic image using non-TOF algorithms by evaluating the effect of number of iterations and subsets(sub-iterations) in the reconstruction process with and without resolution recovery in four different tumour to background ratios (TBRs). We analyzed radial activity concentration profiles for different lesion sizes in PET/CT images and calculate recovery coefficient (RC), contrast to noise ratio (CNR), noise and coefficient of variation(COV) besides activity concentration distribution (Bq/cc) and standardized uptake value (SUV) measurement. Measurements were performed on the Siemens Biograph 6 true points PET/CT scanner (SIEMENS, Erlangen, Germany), using a cylindrical phantom filled with five hot syringes with various diameters (0.5, 0.9, 1.3, 1.6, 2.2cm). The acquisition was done four times with different TBRs (8:1, 6:1, 4:1, 2:1). The images were attenuation corrected using low dose CT images (110kV, 31mA). The reconstruction parameters were iterative, iterative 3D and TrueX (resolution recovery in SIEMENS) reconstruction algorithms with 1-8 iterations and 4,8,14, and 21 subsets in 128x128 matrix size which results in totally 1380 different configurations. It should be noted in all reconstructions, a post smoothing filter with 5mm Full width Half Maximum (FWHM) was used. CNR values were higher than five threshold value, for all the syringe sizes and TBRs except for the syringe size 0.5 in TBR 6:1, 0.9 in TBR4:1, and 2.2 in TBR 2:1. RC Values reported for iterative 2D and iterative 3D has large bias in comparison to TrueX reconstruction algorithm which makes worse in lower number of sub-iterations. Reconstruction parameters can significantly affect SUV and quantification measurements. The optimized reconstruction TrueX algorithm was proposed for PET/CT diagnostic imaging. © 2018. All Rights Reserved.