Abstract
1826
Objectives 18F-FDG PET-CT (18F-fluorodeoxyglucose, positron emission tomography-computerized tomography) is a powerful tool for radiotherapy planning of lung lesions, especially with promising 4D acquisitions. Our aim was to assess a new 4D reconstruction scheme: Multi-RRA (Reconstruct Register and Average) acquired 8-times faster than a standard 4D-PET-CT acquisition. This method was compared to a reference (4D-CT) and to other PET reconstructions (3D-PET and 4D-PET).
Methods PET acquisitions synchronised to the respiratory signal were performed on patients presenting mobile lung lesions. Tumor displacement volumes were compared on each reconstruction (the reference 4D-CT, 3D-PET, 4D‑PET and multi-RRA). Topologic similarity indexes from reference (Dice, Jaccard and Overlap) were also studied.
Results 22 lesions were evaluated. Volumes obtained with multi-RRA were not significantly different from reference (3.70 mL vs. 2.94 mL). Mean bias from reference was low (-0.32 mL) and better than mean bias of the other reconstructions (-0.69mL and -2.5mL for 4D-PET and 3D-PET respectively). No significant difference was found for the displacement volumes between 4D‑PET and multi-RRA. Topologic similarity indexes were slightly better with multi-RRA but not different from those with 4D-PET.
Conclusions Multi-RRA algorithm provides information similar to the one obtained with 4D-PET with a shortened acquisition time.