A managed analysis environment for quantification of neuroimaging studies with microPET

• Published in: The Journal of nuclear medicine (1978) Vol. 58; p. 296
• Main Authors: Nandi, Ayon, Kuwabara, Hiroto, Nakano, Masayoshi, Fedorko, Julie, Zhao, Qian, Roberts, Joshua, Valentine, Heather, Wong, Dean
• Format: Journal Article
• Language: English
• Published: New York Society of Nuclear Medicine 01.05.2017
• Subjects: Alignment; Anesthesia; Cerebellum; Coefficient of variation; Computed tomography; Correlation analysis; Correlation coefficient; Correlation coefficients; Fluorine isotopes; Frames; Glutamic acid receptors (metabotropic); Inspection; Isoflurane; Magnetic resonance imaging; Medical imaging; Mice; Neuroimaging; Neurology; NMR; Nuclear magnetic resonance; Nuclear medicine; Parameters; Pets; Positron emission; Positron emission tomography; Reproducibility; Rodentia; Rodents; Scaling; Scanners; Skull; Straps; Tomography
• ISSN: 0161-5505; 1535-5667

Objectives: Combined PET-CT scanners are widely used for preclinical studies with rodents. However, the combination can create confounding factors in neuroimaging studies because of the lack of anatomical details on CT, and imprecise co-registration of CT and PET volumes. Here we propose and evaluate novel approaches for spatial coregistration of individual CT and PET volumes to a set of standard CT, MRI, and tracer-specific PET volumes. Such approaches, if proven usable, could facilitate use of PET-CT scanners for neuroimaging studies of rodents without restricting head position. Methods: A total of 16 C57/BL6J mice (Jackson Laboratories) were scanned for 60 minutes after IV injection of 200-250 uCi of [18F]-FPEB, a specific tracer for the metabotropic glutamate receptor, type 5 in the Super Argus PET-CT scanner (Sedecal). Two mice were scanned in single sessions under inhallation isoflurane anesthesia (2%) with the mice restricted with straps in 'best-fit' positions (which can lead to crooked head positions). Separately, standard CT, MRI (with VOI template), and [18F]-FPEB PET volumes were prepared1,2. Individual CTs were aligned to the standard CT with initial 6 parameter reorientation (denoted as an upright CT) followed by 3 parameter scaling. Individual PETs were reoriented to their own upright CTs, scaled for individual sizes, and aligned to the standard [18F]-FPEB PET volume. The coregistration and realign modules of SPM12 were used together with manually obtained initial guesses of parameters without which SPM tools failed to converge. Then, dynamic PET frames were transferred to the standard space. Standard VOIs were manually displaced by up-to 2 voxels (0.24 mm) in three directions as needed, and applied to PET frames to generate time activity curves (TACs). Results: On visual inspection, outlines of the skull of the standard CT aligned to normalized individual CTs in all mice for all raters (n=6). Quantitatively, 500 points along the standard skull deviated by less than 0.01 mm from mean positions of raters across mice. For PET, mean deviations ranged up to 0.8 mm among 4 experienced raters. SUV values of 14 standard and [18F]-FPEB-specific regions correlated well among raters (R2>0.996), and showed very low inter-rater variations: Coefficients of variation were less than 1% in all regions except <4% in the cerebellum, and <3% in the hippocampus. The average intraclass correlation coefficient across raters was > 0.9, p < 0.05. Conclusion: The proposed approaches showed excellent inter-rater reproducibility estimates of spatial coregistration for CT (all raters) and PET (if experienced), and of TACs. Thus, the proposed approaches could make combined PET-CT scanners to neuroimaging studies of rodent without strongly restricting head positioning in the scanner.