Truncation artifacts in hybrid PET/MR - evaluation and compensation

Ziel/Aim:

MR-based attenuation correction in hybrid PET/MR imaging is affected by the reduced transaxial FOV of the MR (40cm) in comparison to that of the PET scanner (60cm) which leads to truncation artifacts in the MR-based attenuation map
(MRMap). In this study we investigate the quantitative influence of truncation artifacts on the reconstructed whole-body PET image volumes and present a new method for truncation compensation (TC).

Methodik/Methods:

Data sets of 17 whole-body FDG patients (BMI 16-38) examined with an Philips Ingenuity PET/MR with arms down were reconstructed using different techniques for TC: (1) no TC (PET-noTC), (2) vendor-provided TC (PET-TC1), (3) newly
developed TC (PET-TC2) based on automatic 3D segmentation of PET-noTC. In all PET volumes, the resulting SUVmax/mean of ROIs in different body regions (e.g. heart, bladder, mediastinum, kidneys, arms etc.) were compared.
Furthermore, a correlation analysis between all voxels in the trunk and truncated arms was performed.

Ergebnisse/Results:

Residual artifacts in the truncation compensated MRMaps were found in 11 (TC1)and 3 (TC2) out of 17 cases. Average SUVmax differences between PET-noTC and PET-TC1/PET-TC2 were: -2.5%/-3.2% (bladder), -4%/-3.4% (kidney),
-3.2%/-3.5% (heart), -2.8/-3.3% (abdomen), -29%/-37% (arms). The correlation analysis showed average SUV deviations of -5% to -1% and -60% to -30% depending on patients' BMI for all voxels in the trunk and truncated arms, respectively.

Schlussfolgerungen/Conclusions:

In whole-body PET imaging, truncation of the arms in the MRMap leads to an average SUV underestimation of ca. 3% in the trunk. In the truncated areas SUV underestimations in the range of 30-60% can be expected. Truncation
compensation based on PET-noTC is straightforward and improves quantification accuracy distinctly. According to our experience the procedure is also well suited for clinical routine.