Room: AAPM ePoster Library
Purpose: PET quantification depends on accurate attenuation correction (AC) methods, which is a major challenge for PET/MR systems that use MR images to calculate the 511 keV photon attenuation. To evaluate different AC methods, we employed a PET/MR lesion insertion tool that simulates lesions in a patient cohort that has both PET/MR and PET/CT images. These lesions were processed through the PET scanner model, and the CT-based AC (CTAC) served as the ground-truth. The current study assesses standard update value (SUV) accuracy of different PET/MR AC methods using matched contralateral synthetic lesions with the PET/MR lesion insertion tool.
Methods: 71 lesions were identified in 18 pelvic PET/MR patient data and synthetic contralateral lesions were generated. The generated lesions matched the contrast (mean SUV in the original lesion VOI) and size (matched spherical diameter) of the true lesions and were inserted on the same axial slice, contralateral to the original lesion. The CTAC and the MR-based AC (MRAC) were used for the reconstructions, and the SUV error (ratio of the VOI mean of SUV_CTAC to mean of SUV_MRAC) was calculated for all lesions. The matched pairs of the original and the inserted lesions’ error values were divided (error ratio) and analyzed across the lesion pairs.
Results: The error ratio over uptake lesion pairs was unity (mean error ratio 1.004 stdev ±0.020), which indicates that the synthesized lesions are closely matched to the original lesions with similar SUVs errors characteristics from CTAC and MRAC. There were no linear trends when evaluating error ratio versus lesion diameter size (R² = 0.0073).
Conclusion: The lesion insertion tool can simulate uptake in pelvic lesions and their SUV characteristics. This further strengthens the confidence in using this tool for evaluating novel AC methods that have the potential to improve the quantitative accuracy in simultaneous PET/MR systems.