RT Journal Article
SR Electronic
T1 <strong>Identificationand comparisonof image-derived input functions using total-body PET</strong><sup/>
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 520
OP 520
VO 60
IS supplement 1
A1 Elizabeth Li
A1 Simon Cherry
A1 Alice Tarantal
A1 Hongcheng Shi
A1 Shuguang Chen
A1 Pengcheng Hu
A1 Yu Ding
A1 Debin Hu
A1 Ping Zhou
A1 Tianyi Xu
A1 Chao Wang
A1 Terry Jones
A1 Ramsey Badawi
A1 Guobao Wang
YR 2019
UL http://jnm.snmjournals.org/content/60/supplement_1/520.abstract
AB 520Objectives: Full compartmental modeling is vital in dynamic positron emission tomography (PET) for a quantitative understanding of tracer kinetics but requires a blood input function. A noninvasive image-derived input function (IDIF) can replace arterial sampling, but reliable IDIF extraction (i.e. large blood pool) can be challenging with conventional PET due to the limited axial field-of-view (FOV) (e.g. dynamic brain studies). The EXPLORER [1] is a 2-meter long total-body human PET scanner and the miniEXPLORER I [2] is a 45-cm axial FOV PET scanner for nonhuman primates (NHPs). This study reports the first demonstration of the benefit of total-body EXPLORER-based PET imaging for extracting IDIFs for full kinetic analysis. Methods: The left ventricle (LV) IDIF was compared with the carotid IDIFs for 2-tissue compartment modeling (2TCM). For human EXPLORER imaging, a healthy female (56 kg, age 61) was injected with 6.9 mCi [18F]-fluorodeoxyglucose (FDG) and scanned for 60 minutes. The subject gave informed consent under the guidance of the Ethics Board at Zhongshan Hospital (Shanghai, China). Images were reconstructed following the temporal scheme: 61x1s, 30x2s, 20x3s, 12x10s, 50x30s, 15x120s and were corrected for scatter, randoms, and attenuation. For miniEXPLORER imaging, a healthy male rhesus monkey (5.9 kg, age 5) was injected with 2.8 mCi [11C]-raclopride (RAC) and scanned for 90 minutes. Images were reconstructed with the framing of 30x2s, 24x5s, 21x20s, 20x60, 12x300s with all data corrections applied. In each study, time activity curves (TACs) were extracted from manually drawn brain regions of interest (ROIs) and fitted using a 2TCM. In addition to standard parameters vb, K1, k2, k3, k4 (k4 =0 for FDG), blood time delay was also fit. Akaike information criterion (AIC) of TAC fits and parameter estimates for FDG influx rate Ki=K1k3/(k2+k3) and RAC D2/D3 receptor binding potential (BPND=k3/k4) were used to assess the impact of IDIF choice on 2TCM. Results: Fig. 1A-1D are total-body dynamic images from the EXPLORER study. FDG parameters (K1 and Ki) and AIC are shown in Fig. 1E. K1 and Ki were increased by 30-50% when a carotid IDIF was used. Smaller K1 values using the LV IDIF can be explained by a higher LV IDIF peak, which also impacted Ki. While we do not have ground truth here, lower AIC values by LV IDIF indicate the superiority of LV over the carotids for fitting brain TACs. Further, fractional blood volume vb of FDG was &gt;10% by carotid IDIFs and 3-5% by the LV IDIF. The latter is more consistent with literature. IDIFs and a brain TAC in Fig. 1F clearly demonstrate the need for delay time correction. Carotid IDIFs have lower peak activities compared to LV IDIF, which is likely due to partial volume effects. A similar pattern occurred in the RAC study (Fig. 1G, 1H). Striatal K1 and BPND estimates by the LV IDIF were lower than those with carotid IDIFs and resulted in values closer to literature. Like the FDG study, the LV IDIF achieved lower AIC than carotid IDIFs. Conclusions: Estimation of kinetic parameters by 2TCM was affected by IDIF choice in human FDG and NHP RAC studies. Statistical fit quality using AIC indicates that the LV IDIF (with delay correction) is more appropriate than the partial-volume limited carotid IDIF for full kinetic analysis of brain data. The results demonstrate a benefit of total-body over conventional PET scanners for brain imaging with image-derived input functions. References: [1] Cherry, S, et al. J Nucl Med 59 (2018): 3-12. [2] Berg, E, et al. J Nucl Med 59 (2018): 993-998. Fig. 1: A-C: A series of 5-second FDG dynamic frames shows the bolus pass through the LV (A), carotid arteries (B), and jugular veins (C). D: A 0-60s FDG summation image was produced to visualize blood vessels. E: EXPLORER estimates for vb, K1, and Ki, and AIC values from four brain ROIs and three IDIFs (LV: left ventricle, LC: left carotid, RC: right carotid). F: 0-90s of the FDG TACs show the LV IDIF delay for a brain ROI. G: 0-300s of the NHP TACs for RAC. H: Striatal RAC estimates of BPND and AIC values.