PT - JOURNAL ARTICLE AU - Meier, Joseph AU - Alves, Vinicius AU - Trout, Andrew AU - Sharp, Susan AU - Abu Ata, Nadeen AU - MacLean, Joseph AU - Spangler-Bickell, Matthew AU - Brady, Samuel TI - <strong>Prospective Evaluation of the Impact of Motion Correction Software in Pediatric Brain 18F-FDG PET Imaging: A move towards reduced use of sedation</strong> DP - 2024 Jun 01 TA - Journal of Nuclear Medicine PG - 241273--241273 VI - 65 IP - supplement 2 4099 - http://jnm.snmjournals.org/content/65/supplement_2/241273.short 4100 - http://jnm.snmjournals.org/content/65/supplement_2/241273.full SO - J Nucl Med2024 Jun 01; 65 AB - 241273 Introduction: Despite advances in PET technology permitting faster, higher-quality imaging, sedation may be required for pediatric brain PET to avoid motion artifacts. GE HealthCare has a research-only motion-correction (MoCo) reconstruction algorithm which, through data-driven event-by-event MoCo list mode reconstruction, can correct rigid motion. We aimed to prospectively quantify the impact of this motion correction software on deliberately motion corrupted pediatric and young adult 18F-FDG PET brain images.Methods: In this prospective study, patients &lt;21 years old underwent a 9-minute research 18F-FDG brain PET after a clinically indicated PET. For the first 6 minutes (0-6), participants lay still. In the next 3 minutes (7-9), participants performed scripted head movements at fixed intervals with each patient performing one type of motion: Intermittent head tilting (up, down), intermittent head-turning (left, right), low sustained sweep of the head (gradual turn over 15-20s), or complex motion (random motion every 10 s). Exams were acquired on a 30 cm axial field of view PET/CT (GE Discovery MI Gen 2). Images were reconstructed using a Bayesian blocked sequential penalized-likelihood algorithm with (MC) or without (no) motion correction (MoCo). Eight reconstructions were created for each patient (Figure 1): (1) Minutes 0-6 (still), no MoCo [no_s_360s], (2) Minutes 0-6 (still), with MoCo [MC_s_360s], (3) Minutes 4-6 (still) combined with minutes 7-9 (motion), no MoCo [no_s+m_360s], (4) Minutes 4-6 (still) with minutes 7-9 (motion), with MoCo [MC_s+m_360s], (5) Minutes 4-6 (still), no MoCo [no_s_180s], (6) Minutes 4-6 (still), with MoCo [MC_s_180s], (7) Minutes 7-9 (motion), no MoCo [no_m_180s], and (8) Minutes 7-9 (motion), with MoCo [MC_m_180s]. All used attenuation correction, Beta of 500, 30 cm FOV and a 384 x 384 matrix. Images were independently rated in random order by 3 observers blinded to reconstruction. Reviewers scored (1) presence of motion artifacts (Y/N) and on a 3-rank Likert scale scored: (2) overall image quality, (3) grey/white matter differentiation, (4) basal ganglia definition, (5) uniformity of cerebral cortical uptake. A 4th observer drew 2D, 5 mm diameter circular regions of interest (one on each hemisphere) in (1) the frontal cortex, (2) frontal white matter, (3) basal ganglia and (4) the cerebellar cortex to measure SUVmean, SUVmax, and SUV standard deviation. Statistical significance was tested with the Fisher, Wilcoxon Rank Sum, and Kruskal Wallis tests.Results: 16 study participants (mean age, 13 years ± 3 [SD]; age range, 8–18 years; 8 male, 67.1 ± 24.5 kg) were recruited, and received 0.12 ± 0.001 mCi/kg 18F-FDG. Qualitative results are summarized in Figure 2. All still images with MoCo were scored significantly higher than still images without MoCo, suggesting that involuntary motion blur was corrected. No statistical differences (NSD) in scores were seen when comparing no_s_180 with MC_m_180, suggesting that MoCo corrected image blur even in shorter scans. For full data (360s), differences were only noted between no_s_360 with MC_s+m_360 for grey-white matter differentiation, basal ganglia definition, and cortical uniformity, where MoCo images had higher (better) scores (Figure 1).Quantitatively, NSD was observed for frontal cortex SUVmax and SUVmean between no_s_180 and MC_m_180(p&gt;0.05) as well as n_s_360 and MC_s+m_360 (P&gt;0.05). NSD was observed for frontal white matter SUVmean between MC_s+m_360 and no_s_360 (p&gt;0.05). NSD was observed for SUVmean and SUVmax in the basal ganglia between MC_s+m_360 and no_s_360 (p&gt;0.05). NSD was observed for SUVmax and SUVmean measures of the cerebellum between no_s_180 and MC_m_180 (p&gt;0.05) as well as n_s_360 and MC_s+m_360 (p&gt;0.05).Conclusions: In children and adolescents with scripted motion-corrupted 18F-FDG-PET of the brain, motion correction software produced images that are qualitatively and quantitatively indistinguishable or better than images obtained without motion.