Motion tracking of low‐activity fiducial markers using adaptive region of interest with list‐mode positron emission tomography

• Published in: Medical physics (Lancaster) Vol. 47; no. 8; pp. 3402 - 3414
• Main Authors: Chamberland, Marc J. P., deKemp, Robert A., Xu, Tong
• Format: Journal Article
• Language: English
• Published: United States 01-08-2020
• Subjects: Algorithms; Fiducial Markers; Humans; list‐mode; Motion; motion correction; motion tracking; PET; Phantoms, Imaging; Positron-Emission Tomography; Tomography, X-Ray Computed; motion tracking; motion correction; list-mode; PET
• ISSN: 0094-2405; 2473-4209
• DOI: 10.1002/mp.14206

Purpose
Motion compensated positron emission tomography (PET) imaging requires detecting and monitoring of patient body motion. We developed a semiautomatic list‐mode method to track the three‐dimensional (3D) motion of fiducial positron‐emitting markers during PET imaging.
Methods
A previously developed motion tracking method using positron‐emitting markers (PeTrack) was enhanced to work with PET imaging. A novel combination of filtering methods was developed to reject physiological tracer background, which would drown out the events from the marker if unfiltered. The most critical filter rejects events whose line‐of‐response (LOR) is outside an adaptive region of interest (ADROI). The size of ROI was optimized by exploiting the distinct differences between the distributions of events from background and marker. The ADROI PeTrack method was evaluated with Monte Carlo and phantom studies. A 92.5‐kBq
22
Na marker moving sinusoidally in 3D was simulated with Monte Carlo methods. The simulated events were combined with list‐mode data from cardiac PET imaging patients to evaluate the performance of the tracking. In phantom studies, three
22
Na markers were placed on a dynamic torso phantom with an initial activity of 680 MBq of
82
Rb in its cardiac insert. The motion of the markers was tracked while the phantom simulated various types of patient motion. Motion correction on an event‐by‐event basis of the list‐mode data was then applied and images were reconstructed.
Results
Simulation results show that the background rejection methods can significantly suppress the tracer background and increase the fraction of marker events by a factor of up to 2500. A 92.5‐kBq marker can be tracked in 3D at a frequency of 2.0 Hz with an accuracy of 0.8 mm and a precision of 0.3 mm. The phantom study experimentally confirms that the algorithm can track various types of motion. The relative accuracy of the experimental tracking is 0.26 ± 0.14 mm. Motion‐corrected images from the phantom study show reduced blurring.
Conclusions
An algorithm and background rejection methods were developed that can track the 3D motion of low‐activity positron‐emitting markers during PET imaging. The motion information may be used for motion‐compensated PET imaging.