The effect of susceptibility artifacts related to metal implants on adjacent lesion assessment in simultaneous TOF PET/MR

Abstract

OBJECTIVE Metal implants could affect the attenuation correction (AC) in positron emission tomography (PET)/magnetic resonance (MR) imaging. The purpose of this study is to evaluate the effect of susceptibility artifacts related to metal implants on adjacent metabolically active lesions in clinical simultaneous PET/MR scanning for both time-of-flight (TOF) and non-TOF reconstructed PET images.
METHODS We included 27 patients without implants, but with confirmed 2-deoxy-2-(18)F-fluoro-D-glucose ((18)F-FDG) avid lesions adjacent to common implant locations. In all patients, a clinically-indicated whole-body (18)F-FDG PET/MR was acquired. Baseline non-TOF and TOF PET images were reconstructed. Reconstructions were repeated after the introduction of artificial signal voids in the AC map to simulate metal implants in standard anatomical areas. All reconstructed images were qualitatively and quantitatively assessed and compared with the baseline images.
RESULTS A total of 51 lesions were assessed. In 40/50 (non-TOF/TOF) of these cases the detectability of the lesions did not change, in 9/1 cases the detectability changed and in 2 non-TOF cases the lesions were no longer visible after the introduction of metal artifacts. The inclusion of TOF information significantly reduced the artifacts due to simulated implants in the femoral head, sternum, and spinal locations (P = 0.01, 0.01, 0.03, respectively). It also improved the image quality in those locations (P = 0.02, 0.01, 0.01, respectively). The mean percentage error was -3.5% for TOF and -4.8% for non-TOF reconstructions, meaning that the inclusion of TOF reduced the percentage error in maximum standardized uptake value (SUVmax) with 28.5% (p<0.01).
CONCLUSION Qualitatively there was a significant reduction of artefacts in the femoral head, sternum, and the spine. There was also a significant qualitative improvement in image quality in those locations. Furthermore, our study indicates that (simulated) susceptibility artifacts related to metal implants have a significant effect on small and moderately (18)F-FDG avid lesions located close to the implant site which possibly may go unnoticed without TOF-information. On larger, high (18)F-FDG avid lesions, the metal implants had only limited effect. The largest and significant quantitative difference was found in artefacts of the sternum. There was only a weak inverse correlation between lesions affected by artefacts and distance from the implant.