A new algorithm (MotionFreeze-MF) for respiratory motion management in PET/CT scan: Comparison to respiratory gated PET/CT (4D) data

Objectives: Aim of the work was to compare respiratory MF corrected images to 4D images as standard reference in lung and liver lesions. Methods: 31 lesions (18 lung and 13 liver) in 28 patients submitted to 4Dscan (Discovery 600-GE Healthcare - Milwaukee; WI-US) from June 2011 to October 2011, were retrospectively reprocessed and analysed with MF. MF performs a non rigid coregistration of all the 4D bins (6 bins; 1.5 min/bin; 9 min FOV duration) to obtain motion corrected static images. Using 4D datasets, MF images with FOV duration of 9 min (9MF), 4.5 min (4.5MF) and 3 min (3MF) were reconstructed. 4D and MF images were visually analysed separately by three blinded nuclear medicine physicians. Lesions were scored positive, equivocal or negative to assess lesion detectability. Image quality was visually scored from 1 (bad) to 5 (very good). MF data were compared to 4D ones as standard reference. Results: 4D, 9MF and 4.5MF findings were concordant for lesion detectability in 29/31 (93.5%) cases; regarding discordant results, 1/31 (3.2%) 4D negative lesion became positive on MF and 1/31 (3.2%) 4D positive lesion became equivocal. 4D and 3MF were concordant in 25/31 (80.6%) lesions and discordant in 6/31 (19.3%). Particularly, 1/31 (3.2%) 4D negative lesion became equivocal, 3/31 (9.7%) 4D positive lesions shifted to equivocal, 1/31 (3.2%) 4D positive changed to negative and 1/31 (3.2%) 4D equivocal changed to negative in MF. The mean +/- SD image quality scores of 9MF, 4.5MF, 3MF and 4D datasets were 4.97 ± 0.18, 4.68 ± 0.54, 3.94 ± 0.36, 4.16 ± 0.58. Conclusions: 9MF and 4.5MF images show comparable lesion detectability and image quality. In particular 4.5MF can obtain similar results of 4Dscan with a shorter acquisition time (4.5 min/FOV vs 9 min/FOV, respectively)