Investigating Low-Dose Image Quality In Pediatric Tof-Pet/Mri

304 Objectives: This work investigates the image quality of FDG-PET for pediatric patients as the injected activity is reduced, lowering the radiation dose to the patient, when using a new high-sensitivity, time-of-flight (TOF) PET/MRI scanner. Avoiding unnecessary radiation exposure is of utmost importance when imaging pediatric patients, as their sensitivity to radiation is heightened, and for patients cured of their disease, there is a greater risk for radiation-induced cancers because of longer life expectancies. The aims of this work were to characterize the quantitative accuracy of FDG uptake for simulated low-dose imaging studies and investigate the improvement with TOF-PET reconstruction for pediatric patients. The results will be used to generate new imaging protocols and injected activity regimens that cause less radiation dose, which ultimately can be compared with PET/CT to evaluate the benefits of TOF-PET/MR in pediatrics. Methods: A combination of phantom imaging and patient data was used in the analysis. In this retrospective IRB approved study between 12/2016 - 12/2017, whole-body FDG scans of pediatric patients with metastatic liver lesions were selected for analysis. The range of patient ages included in the study was 4 to 18 yr-old. The PET/MRI protocol used included T1, T2, and diffusion weighted MRI acquisitions, and a 3 min PET acquisition per bed position. Liver lesions where identified using the clinical images at an injected activity regime of 3.7 MBq/kg; simulated low-dose images were created by removing events from the PET list file in post processing, and were evaluated down to 1/9th dose (0.4 MBq/kg). Several imaging metrics were quantified: bias and variability of FDG in healthy liver tissue, and SUV max, mean, and contrast-to-noise ratio (CNR) of liver lesions. A human reader study was also performed using a Likert-scale to rank the quality of images and to count the number of lesions detected on the PET only and PET/MRI fused images. Results: A total of 23 lesions were identified from the set of patient images. The bias and variability of FDG uptake in healthy liver tissue was found to increase in simulated low-dose images for SUV max, however SUV mean was stable down to very low doses. For younger patients (1-5 yr-old), FDG avid liver lesions in 1/2-dose images had a 6% increase in SUV max and a 14% increase in images with 1/3-dose. An opposite trend was seen in older, larger patients, suggesting a size-dependence in the PET scatter correction. There was a wide range of liver lesion CNR measured on full-dose images (values from 53-18). CNR in simulated low-dose images all had significant reductions below 1/2 dose. This trend in image quality was also observed in results from our qualitative reader studies, however our results suggest that confidence in lesion detection from fused low-dose PET and MR images is greatly improved compared to PET alone. Conclusions: Identification of metastatic liver lesions represents one of the greatest challenges for FDG-PET imaging, as the high organ background uptake and abdominal photon attenuation locally reduce image contrast. Our study demonstrates that dose reduction is feasible for pediatric patients without degrading the diagnostic quality when using TOF-PET/MRI. Results show that the quantitative accuracy of SUV uptake is maintained down to 1/3 dose, however the visual image quality noticeably degrades for low-dose images, and based on our reader studies, a maximum dose reduction of 1/2 is suggested.