@article {Bec429, author = {Julien Bec and David Henry and Andre Kyme and Roger Fulton and Ramsey Badawi and Simon Cherry}, title = {Optical motion tracking for use with the EXPLORER total-body PET scanner}, volume = {59}, number = {supplement 1}, pages = {429--429}, year = {2018}, publisher = {Society of Nuclear Medicine}, abstract = {429Objectives: To provide high precision tracking information of the patient position in a total-body positron emission tomography (PET) scanner (EXPLORER). Total-body PET is an exciting development in the field of nuclear medicine. Through the EXPLORER project, a clinical system is being developed and will be ready for patient studies early 2019. Patient motion is a well-identified source of image artifact, especially severe in the pediatric and elderly population and may be exacerbated by the enclosed imaging environment of a total-body PET scanner. This can be addressed by various methods aimed at maximizing comfort, from optimization of thermal conditions, patient positioning, design of specific beds and holders, to in-bore entertainment and patient training prior to imaging. To provide guidance in the development of patient motion limitation methods, we evaluated the performance of a motion tracking system composed of an out-of-bore stereo camera system integrated with a full-size mock-up of the EXPLORER scanner that includes a functional bed system. Methods: The EXPLORER mock-up (United Imaging Healthcare) consists of the front and back covers of a clinical PET/CT scanner with extended tunnel and outside covers, as well as a functional bed system to fully represent the future EXPLORER scanner (2960 mm length, 700 mm CT bore diameter, 760 mm PET bore diameter). The motion capture system consists of 6 cameras (Prime 13, Optitrack) with a synchronization module, passive infrared reflective markers, and dedicated software. The cameras were positioned outside of the mock-up, with three cameras placed at each end of the bore, to provide good line of sight of the bore without interfering with patient loading (Fig. 1a), leading to near collinear lines of sight (Front: ~20 deg/ Back: ~4.5 deg) and thus suboptimal triangulation of the bore. The accuracy of the tracking system in this configuration was evaluated by tracking a marker rotated at a constant speed (40 rpm). The rotation plane was set such that it intersected all three axes of the scanner{\textquoteright}s reference frame. The component of this trajectory for each axis is expected to be purely sinusoidal, and therefore the positional accuracy could be assessed by calculating the residuals of a sinusoidal fit of data provided by the motion capture system (Fig. 1b). The accuracy measurement was performed at various locations in the bore, using all cameras as well as cameras from a single side at a time to simulate visual obstructions (Fig. 1c). Subsequently, markers were positioned inside of the bore, on the bed and on the chest of a human subject and were used to evaluate chest motion during normal breathing and breath-hold. Results: Fig. 1b shows the steps needed to perform accuracy evaluation with the proposed method. A typical error distribution for the horizontal axis (across bore) is displayed. Fig. 1c show that high accuracy measurements can be performed over the whole bore volume in spite of suboptimal camera positioning. Using cameras from one side only, the positioning accuracy degrade as the distance to the target increases. For vertical and lateral motion, accuracy is very high with less than 0.25 mm average error in all configurations. As expected, positioning accuracy is lowest along the bore axis. It can be noted that the lowest performances in the Z direction are obtained using only the rear cameras, placed close to each other. Conclusion: We demonstrated that an optical motion tracking system could be implemented to track sub-millimeter motion in a full body PET/CT geometry despite severe limitations in camera placement due to the scanner geometry. A quick and efficient way to estimate positioning accuracy was devised so that accuracy could be characterized over the bore volume and in the presence of obstructions. The positioning accuracy achieved here is suitable for the stated goal of evaluating strategies to limit patient motion. Acknowledgements: This work was funded in part by NIH grant R01CA206187}, issn = {0161-5505}, URL = {https://jnm.snmjournals.org/content/59/supplement_1/429}, eprint = {https://jnm.snmjournals.org/content}, journal = {Journal of Nuclear Medicine} }