The optimization of 4D PET acquisition parameters for radiation oncology

Objectives The use of 4D PET for mobile tumors is of great interest to radiation oncologists when they try to treat with gating or using an internal target volume approach for radiation therapy planning. We have measured a PET phantom on a motion table to assess the potential distortions.

Methods A Jaszczak phantom with hollow spheres of various volumes (0.3 - 14cc) was filled with F-18 water using 3 different sphere-to background ratios (SBR), ranging from 3.5:1 to 10:1. Background was 0.07mCi/Kg approximating the average activity in a radiation oncology patient at the time of imaging. The phantom was mounted on a motion table programmed to the move the phantom in 3 different motion patterns; 1 cm, 1.5 cm and 2.0 cm all with a respiratory cycle of 5 seconds. The volumes as determined by software were compared to the actual volume as determined from the CT. SBR values were also compared to measured values for all sphere sizes. To assess SBR accuracy the software used a SBR threshold of 50% of SUV max.

Results The biggest variation in ration of PET/CT volumes was observed when using 2 cm motion and using 10 phases for binning the PET volume, with a volume variation of the largest sphere of 7%, while the variation of maximum SBR varied 11.7%. When the 6 phase binning was used the variation seen in volume was reduced to 3% and the maximum SUV variation in this case was 7%. The largest motion pattern yielded the biggest variation in both PET/CT volumes and SUV variation. For 1 cm motion the volume variation was reduced to 5.3% and SUV variation was reduced to 5%. The smallest sphere was not observable in all phases binned with 6 or 10 phases. However this problem was observed in fewer phases using the 6 phase technique.

Conclusions The use of 6 phases reduces the distortions with all volumes and provides superior detection for small lesions. Using the 256 X 256 matrix with 5 mm FWHM filter provided most accurate SBR and improved the small volume detection