Motion Correction for Respiratory Gated PET Images

The aim of this study was to investigate the feasibility of a motion correction method for respiratory gated PET and to evaluate its effect on image quantification. The NCAT phantom was used and simulated to have three lung lesions at different locations in the right lung. All lesions had either a source to background ratio of 5:1 or 10:1. Data were binned into 16, 8, 4, and 2 gates in addition to a non-gated data set. Poisson noise was added to the data before being reconstructed with OSEM, 4 iterations and 8 subsets. Using a non-rigid registration algorithm, the gated images were deformed into the peak inhale gate. This resulted in the motion corrected images, which were then summed for analysis. Regions of interest were placed in the lung background, soft tissue, and the center of each lesion. The mean signal, contrast to noise ratio (CNR), and spatial resolution were evaluated as a function of the number of gates, with and without motion correction, lesion size, lesion placement, lesion contrast, and count level. Compared to the non-gated data, mean signal recovery in the gated and motion corrected images increased as the number of gates used increased, lesion placement was lower in the lung, as the lesion size increased, and as lesion contrast decreased. Although the mean value did not change significantly as the number of gates increased from 4 to 16, the standard deviation decreased significantly. This resulted in an increase in CNR recovery particularly with decreased lesion size, lower lesion placement, and decreased intrinsic contrast. The spatial resolution also improved 10-30% in the motion corrected images for the lesions located at the lower and middle of the lung. These results indicated that the motion corrected images have greatly reduced image noise, and at the same time improved signal recovery due to reduced resolution losses. This showed the efficacy of this motion correction algorithm, and resulted in improved image contrast and detectability.