Correction of an image size difference between positron emission tomography (PET) and computed tomography (CT) improves image fusion of dedicated PET and CT

Abstract

Clinical work in software positron emission tomography/computed tomography (PET/CT) image fusion has raised suspicion that the image sizes of PET and CT differ slightly from each other, thus rendering the images suboptimal for image fusion. The aim of this study was to evaluate the extent of the relative image size difference between PET and CT and the impact of the correction of this difference on the accuracy of image fusion. The difference in real image size between PET and CT was evaluated using a phantom study. Subsequently, 13 patients with cancer in the head/neck area underwent both CT and [(18)F]fluorodeoxyglucose PET in a custom-made mask for external beam radiotherapy, with multimodality markers for positional reference. The image size of PET relative to CT was determined by evaluating the distances between the markers in multiple directions in both scans. Rigid-body image fusion was performed using the markers as landmarks, with and without correction of the calculated image size difference. Phantom studies confirmed a difference in real image size between PET and CT, caused by an absolute error in PET image size calibration. The clinical scans demonstrated an average relative difference in image size of 2.0% in the transverse plane and 0.8% along the longitudinal axis, the PET images being significantly smaller. Image fusion using original images demonstrated an average registration error of 2.7 mm. This error was decreased to 1.4 mm after size correction of the PET images, a significant improvement of 48% (P<0.001). A significant deviation in PET image size may occur, either as a real image size deviation or as a relative difference from CT. Although possibly not clinically relevant in normal diagnostic procedures, correction of such a difference benefits image fusion accuracy. Therefore, it is advisable to calibrate the PET image size relative to CT before performing high-accuracy rigid-body image fusion.