Paper

Investigation of Reconstruction Algorithms by Means of Quantitative PET/CT Phantom Studies with a Focus on Evaluation Strategies A PET/CT Phantom Study


Authors:
Thomas Layer; Matthias Blaickner; Wolfgang Birkfellner; Anton Staudenherz; Johannes Neuwirth
Abstract
The aim of this work is to investigate the TrueX reconstruction algorithm and to compare different PET reconstruction algorithms (2D OSEM and 3D OSEM) using a phantom with non-spherical objects. Furthermore, different evaluation algorithms and their impact on activity analysis are explored. Measurements were carried out using a Siemens Biograph 64 TruePoint™ PET/CT Scanner and an in-house phantom. The measurements were carried out at four different signal to background ratios (SBR), using 18F-FDG. For evaluation purposes, three different algorithms were used: Adaptive Thresholding, Percentage Thresholding, and the Maximum Line Method. With regard to the two thresholding procedures, the effect of subtracting the voxels of the inactive plastic walls was investigated. TrueX produces rather large overestimates of the activity concentration of up to 25 % at diameters between 11 mm and 24 mm. The technique of stripping off the inactive hull after Adaptive and Percentage Thresholding considerably reduces the contribution of the partial volume effect, thus minimizing the underestimate, which is much more constant over the entire range of diameters. The thresholding procedures generate statistically stable results, but their detection efficiency decreases with decreasing diameter and SBR. The Maximum Line Method naturally yields results for all cylinders and parameter settings; however, this occurs at the cost of a higher relative uncertainty. For quantitative studies, such as quality assurance measurements with standard phantoms, 3D OSEM should be preferred.
Keywords
TrueX; PET-CT; Activity Analysis; OSEM
StartPage
7
EndPage
15
Doi
10.5963/LSMR0502001
Download | Back to Issue| Archive