Abstract
3086
Objectives: Image artifacts may be generated by indwelling urinary catheters to appear outside the body in 18F-FDG PET/CT imaging. It has been inferred that the cause of such artifacts is scatter correction errors, overestimates of the scatter contribution causing the halo-artifacts, and hindering evaluations of tumor. The purpose of this study is to investigate the incidence and mechanism of the halo-artifact generation, and to establish a feasible method to achieve corrected scattering to be able to reduce these halo-artifacts.
Methods: The study reports patient and phantom studies, with the patient studies, based on 18F-FDG PET/CT images of patients with indwelling urinary catheters that were retrospectively evaluated. Using axial PET images depicting the urinary catheters, PET images before and after scatter correction were compared to evaluate the presence or absence of halo-artifacts. Further, to determine the incidence of the artifacts, the volume-of-interest was set to the whole body and the average standardized uptake value (SUV) was measured for axial PET images depicting the urinary catheters. In the Phantom studies, using real urinary catheters and a NEMA body phantom, the appearance pattern of halo-artifacts presumed in Patient studies was reproduced and verified. The scatter correction method was a tail-fitting scaling with single scatter simulation (TF-SSS) (TFS 4 mm) using the default setting of a 4 mm voxel μ-map. Further, images by TF-SSS (TFS 2 mm) and Monte-Carlo simulation scaling with single scatter simulation (MC-SSS) using a μ-map of 2 mm voxels were also used in the study.
Results: In patient studies, of the 3,633 cases here, there were 59 patients with indwelling urinary catheters. In 8 (14%) of the 59 cases, halo-artifacts were observed in the TFS 4 mm images.(Fig. A) In 4 of the 5 cases, the halo-artifact was not present in the TFS 2 mm images.(Fig. B) However, in all 5 cases, halo-artifacts were absent in images that were not subjected to scatter correction and also absent in the MC-SSS images.(Fig. C) The SUV was 0.5 ± 0.2 for TFS 4 mm, 0.7 ± 0.2 for TFS 2 mm, and 0.9 ± 0.3 for the MC-SSS images. The Phantom studies reproduced the appearance of the halo-artifacts. The TF-SSS image underestimated the SUV at the site where the halo-artifact was present, and the value shown by the MC-SSS image was close to the actual value. In the TF-SSS images, halo-artifacts were caused by overestimates of the scatter components due to mismatches between the μ-map and PET images. This mismatch was presumed to be caused by urine movement in the interval between the CT and PET imaging, or by disappearance of the urinary catheter on the μ-map due to the partial volume effect and smoothing in the processing of images.
Conclusions: The cause of the halo-artifacts is a mismatch between the μ-map and PET, which induces errors in the scatter correction. With the TF-SSS method the halo-artifact was improved when reducing the μ-map voxel size, but the artifacts remained in some cases. With the MC-SSS method, it was possible to conduct an accurate scatter correction and to eliminate the halo-artifacts.