PT  - JOURNAL ARTICLE
AU  - Zhang, Xuezhu
AU  - Xie, Zhaoheng
AU  - Cheng, Zhaoping
AU  - Duan, Yanhua
AU  - Xu, Tianyi
AU  - Li, Chenwei
AU  - Li, Hongdi
AU  - Jones, Terry
AU  - Cherry, Simon
AU  - Badawi, Ramsey
AU  - Qi, Jinyi
TI  - <strong>Real-time Motion Tracking for Total-body Molecular Imaging</strong>
DP  - 2022 Aug 01
TA  - Journal of Nuclear Medicine
PG  - 2601--2601
VI  - 63
IP  - supplement 2
4099  - http://jnm.snmjournals.org/content/63/supplement_2/2601.short
4100  - http://jnm.snmjournals.org/content/63/supplement_2/2601.full
SO  - J Nucl Med2022 Aug 01; 63
AB  - 2601 Introduction: Introduction: Total-body molecular imaging can provide physiologically and biologically relevant tracer kinetics across the entire living subject. Although total-body PET, through its high sensitivity, is able to offer high image quality, object motion remains a degrading factor leading to reduced detection and quantification accuracy for clinical research, disease diagnosis and treatment monitoring. Current data-driven approaches require complex post-processing and cannot obtain real-time motion signal in PET scans, which hinder the timely quantitation of motion entangled high temporal resolution dynamic imaging. To overcome this limitation, we develop a real-time singles-based motion tracking approach for correction of total-body PET data and present the initial results. Methods: Methods: Total-body PET data acquired on uEXPLORER PET/CT was used in this study. The motion tracking signal was derived from singles rate measurement. Unlike previous studies that used raw counts which did not work well, we separate the singles rates based on the axial position to take advantage of the total-body coverage of the uEXPLORER. The baseline variation caused by tracer redistribution was first extracted using a Gaussian filter (σ=10 s) and removed from the singles rate curves. A weighted sum of the singles rates of different detector rings was obtained by the strength and phase of the motion signal that they possess to improve the signal-to-noise ratio. The weighted sum singles rate curve is further processed by a low-pass filter with a cutoff frequency of 0.5 Hz to remove high-frequency noise that is unrelated to respiratory motion. The resulting motion signal can be used to perform either phase- or amplitude-based gating. To test the effectiveness of this method, we recruited lung cancer patients with regular and irregular respiratory rhythms for 1-hour 18F-FDG scan, who were given informed consent under the guidance of the Ethics Board of Shandong Qianfoshan Hospital. The latter case includes a mixture of deep breath, shallow breath, and short breath cycles. An external breathing belt was placed on the patients to obtain reference respiratory signal. We evaluated the performance of the proposed singles-based method by identifying the end-inspiration peaks and calculating the timing error with respect to those from the reference signal. We reconstructed the dynamic images on extracted motion gates, and compared them in ungated and gated states.Results: Results: We found that motion signal can be observed in the raw singles and coincidence counts rate curves, where the singles rate curve provides higher SNR over those based upon the prompts and randoms rates. The detector ring covering the abdominal region provides the strongest respiratory motion signal, while the detector ring covering the brain also provides signal but with an inverted phase relative to the regions under the lungs. The SNR from the ring covering the thoracic region is relatively weak because of the phase cancellation of signals from the organs moving inside the chest. The singles rates of different rings offer varied signal strength and noise levels. The SNR ratio of the motion signal is substantially improved by the weighted sum of singles rate over the raw singles rate. The percentage of matched end-inspiration peaks (within ± 1s) between the breathing belt signal and the singles-derived signal was 98.5% for the regular respiratory case and 95.8% for the irregular respiratory case. Among the matched peaks, the timing errors (mean ± s.d.) are 0.01 ± 0.18 s, indicating that the proposed motion tracking method is fairly accurate. The reconstructed gated dynamic images show high image quality.Conclusions: Conclusion: We have developed a singles-based method of real-time motion tracking for total-body dynamic PET imaging. The proposed method does not require complex processing associated with existing data-driven motion detection methods and can be integrated as a built-in component in PET scanners.