Abstract
1714
Objectives: Respiratory motion in static whole body(SWB) PET/CT results in underestimations of measured lesion activity concentration. Phase based gating(PBG) divides the pet data into multiple phases for each respiratory cycle, reducing blur but increasing noise. Patients tend to spend more time in the quiescent phase of the breathing cycle, but to varying degrees. General Electric has implemented quiescent phase gating as the Q.Static algorithm which selects data from each respiratory cycle with a default phase offset(typically 35%) and window width(typically 40%) resulting in a selection of a fixed phase range from 35-75%. This work investigates the effects of personalizing the selection of the phase offset and window width compared to the default settings for a cohort of patients impacted by respiratory motion.
Methods: 10 patients acquired using a 4D PET/CT research protocol, having lung or liver lesions less than 3cm in diameter and impacted by respiratory motion were analyzed in this study. All patients were scanned on a GE Discovery 690 PET/CT. A 12 minute list mode acquisition with respiratory gating using the Varian RPM was acquired over the region impacted by respiratory motion. Triggers were retrospectively inserted into the LIST file at the 0% phase for all respiratory cycles. A 2-10s trigger to trigger acceptance widow was applied. QStatic data was then unlisted using 5 windows widths of 20, 25, 30, 35 and 40 %. For each window width, 8 phase offsets were used (29, 32, 35, 38, 41, 44, 47, and 50). Additionally, for each respiratory cycle we unlisted the data by selecting the ideal offset (OFS ideal) based on the least integral amplitude for each of the 5 window widths. For all cases, the length of unlisted data was changed to keep similar counting statistics across all window widths(12 min for 20% window, 6 min for 40% window). Attenuation correction of the PET data was done using the 56% percent phase from 4DCT for all reconstructions. For all reconstructions, 2it 24ss 6.4mm, TOF, PSF, and 192 x 192 matrix was used. All reconstructions were performed on the GE Discovery 710 PET/CT using QStatic. The SUVmax and SUVpeak of one lesion was analyzed for each of the ten patients. The SUV ratio (max and peak) of the various selections compared to the default QStatic parameters (OFS_4035) were calculated and then the maximum value for each window width and offset combination (OFSmax) were determined. For each window width, for all tumors, the average of the phase offset(not including OFS_ideal) having the highest SUV ratio was calculated . For each window width, the number of times that OFS_ideal resulted in the highest SUV ratio was determined.
Results: The average SUVmax ratios of OFSmax/OFS_4035 were 1.04, 1.08, 1.10, 1.12, and 1.13, for window widths of 40, 35, 30, 25, and 20 respectively. The average SUVpeak ratios of OFSmax/OFS_4035 were 1.02, 1.04, 1.05, 1.06, and 1.07 for window widths of 40, 35, 30, 25, and 20 respectively. The average offset that had the highest SUVmax ratios of OFSmax/OFS_4035 were 31.5, 34.6, 38.6, 35.8, and 43.5 for window widths of 40, 35, 30, 25, and 20 respectively. The average offset that had the highest SUVpeak ratios of OFSmax/OFS_4035 was 31.8, 36.5, 39.2, 40.1, and 42.9 for window widths of 40, 35, 30, 25, and 20 respectively. For SUVmax, the number of times the OFS ideal resulted in the highest ratios of OFSmax/OFS_4035 were 4, 3, 5, 5, and 3 times for window widths of 40, 35, 30, 25, and 20 respectively. For SUVpeak, OFS ideal resulted in the highest ratios of OFSmax/OFS_4035 for 3, 5, 6, 6, and 5 times for window widths of 40, 35, 30, 25, and 20 respectively.
Conclusions: Personalization of quiescent phase parameter selection variably impacts tumor quantification depending on patient breathing waveform. Decreasing the window width had the largest impact on tumor quantification. Our work demonstrates in the majority of cases using a variable offset which is optimized to each patients individual respiratory cycles is superior to using a fixed phase offset
In this issue
