Abstract
1804
Objectives Hardware based respiratory gating has been offered as a solution to the image degradation caused by patient motion. This strategy can be used to delineate the phases of respiratory motion, thus improving resolution, but requires additional time, cost and effort. In the last few years data driven (software) gating - a strategy based on extracting motion information directly from raw PET data - has been developed and refined to run with increasing accuracy and robustness. Data driven gating has the advantages that it is fast, fully automated, reproducible, and requires no changes to clinical acquisition procedures. In this work we compare modern hardware vs. software gating strategies in a large population of PET data.
Methods 189 FDG PET scans were acquired (in 116 patients). Acquisitions were reconstructed as respiratory gated images using hardware derived respiratory triggers (pressure belt) and software derived signal (fully automated post processing method previously presented at SNMMI). Gated images were used to generate gif animation loops of the respiratory cycle, and presented to a nuclear medicine physician for analysis. The only potential difference between the two sets was clarity of motion. Viewing the two animations side-by-side in a randomized order, the observer was asked to delineate which image set had better quality or if they were equal.
Results Randomized review showed that the hardware and software performed comparably. In 93% of the scans motion quality was determined to be indistinguishable. 1% of the hardware and 6% of the software gated scans exhibited optimal (in the set) motion quality.
Conclusions We have shown that data driven gating can work comparably to hardware in a large FDG PET population, and in some cases outperforming it. Our results, and the obvious practical advantages of data driven gating, further the contention that software gating could be a viable strategy for overcoming the challenges posed by respiratory motion and implemented in a robust and ubiquitous manner.
Research Support We would like to acknowledge Siemens AG for their support.