Clinical validation of data-driven PET respiratory gating: impact on image quality and lesion conspicuity

Objectives: Data-driven PET respiratory gating (DDG) technology allows for better integration of respiratory gating into PET clinical workflows by eliminating the need for equipment setup and by enabling gating to be applied retrospectively. DDG methods, however, require clinical validation as they rely on indirect measurements of the respiratory waveforms extracted from list mode data. This study evaluated the clinical impact of a new commercial DDG method on PET image quality and lesion conspicuity in comparison to current standard of care imaging.

Methods: Subjects undergoing clinical ¹⁸F-FDG PET/CT imaging were prospectively enrolled in the study. Subjects were eligible for the study if prior imaging showed evidence of lesions that might be better evaluated with respiratory gating as determined by an experienced nuclear medicine physician. This included subjects with suspicious lesions in the liver, mid-to-lower lungs, or mediastinum. PET/CT imaging was performed on a GE Discovery PET/CT 710 time-of-flight scanner. For comparison, the Varian Real-Time Position Management (RPM) external tracking system was used during PET acquisition. Acquisition times for bed positions over the lungs and liver were extended to 6 minutes instead of the standard clinical 3 minutes. GE Healthcare’s MotionFree (MF) software was used to extract respiratory waveforms from list mode data for each subject. Two sets of motion-corrected images were created by reconstructing PET data acquired during the quiescent phase (50% of the waveform) of both the MF and RPM waveforms. An additional time-matched static non-gated series (3 min) was reconstructed for comparison. Two blinded experienced nuclear medicine physicians independently reviewed the 3 series side-by-side for each subject and rated the image quality on a 5-point scale. Readers also determined if any lesions or lesion groups were more or less conspicuous in any of the 3 series.

Results: Forty subjects were enrolled in the study and completed imaging. A total of 83 lesion groups were evaluated, with a median of 2 (range: 1-6) lesion groups per subject. In the evaluation of overall image quality, both physicians found the MF, RPM, and non-gated series to be equivalent for a large majority of the cases. The MF series was preferred over the non-gated series in 3/40 cases and in 1/40 cases according to physician 1 and physician 2, respectively. MF images were preferred over RPM images in 3/40 (physician 1) and 4/40 (physician 2) cases. In 1 case, physician 1 favored RPM and non-gated images over MF images. In the evaluation of lesion conspicuity, physicians also found the three series to be mostly equivalent. MF improved conspicuity in 1/83 (physician 1) and 4/83 (physician 2) lesion groups when compared to non-gated imaging, and in 1/83 (physician 1) and 7/83 (physician 2) lesion groups when compared to RPM. For 1 posterior lung lesion, non-gated imaging produced better conspicuity than MF and RPM according to physician 2. RPM created artifacts for 2 subjects due to the RPM system temporarily losing track of the external fiducial, which resulted in low-count noisy bed positions. There were no cases in which a lesion was visible in one series but not in the other series.

Conclusions: For a majority of the subjects, the image quality and lesion conspicuity for the MF, RPM, and static series were clinically equivalent. Overall, MF images were not inferior to, and in some cases better than, current standard of care imaging. Despite the low frequency of lesions that benefited from respiratory gating, the ability to retrospectively apply DDG may prove to be important in complicated clinical cases. Acknowledgements: This work was supported by GE Healthcare