Subjects
This study was approved by the Medical Ethics Committee of Zhongshan Hospital, Fudan University, and inform consented was obtained. Ten healthy volunteers (7 men, 3 women; mean age, 55.7 ±9.5 years, mean ±standard deviation) without claustrophobia participated in this study. The average height and weight of the male subjects were 166.9 ±3.7 cm (mean ±standard deviation) and 63.6 ± 8.6 kg, respectively, and for female subjects the average height and weight were 153.7±7.5 cm and 56.2±7.4 kg. Before performing the PET/CT imaging, all subjects were asked to refrain from any medication and to fast at least for 6 h, and their blood glucose was measured by blood sampling.
Total-body PET scanning protocol
A 75-min duration scan was performed immediately after an intravenous injection of F-18 FDG via a vein near the ankle; a dosing regimen of 1.85 MBq/kg was used. The list-mode PET data was acquired on a 194-cm long axial FOV (an axial acceptance angle of ∼± 57°) total-body PET/CT (uEXPLORER, United Imaging Healthcare, Shanghai, China) [21]. Low-dose CT scans were obtained for attenuation correction and all corrections were applied to the reconstructed images. A high temporal resolution dynamic reconstruction protocol was used for the initial 75-min duration scan, with the dataset divided into 60 frames: 36 × 5 s, 24 × 180 s. Another three PET scans of 15-min duration were conducted for each subject approximately at 150, 300, and 480 min after the intravenous injection; these three delayed scans used a dynamic reconstruction protocol of 5 frames: 5 × 180 s. Example reconstructed dynamic PET images in 4 periods for a single subject is shown in Fig. 1. All PET images were reconstructed using ordered subset expectation maximization (OSEM) algorithm with the following parameters: TOF and PSF modeling, 2 iterations and 20 subsets, matrix 192 × 192, slice thickness 2.89 mm, FOV 600 mm (pixel size 3.125 × 3.125 × 2.89 mm3) with a Gaussian post-filter (3 mm), and attenuation and scatter correction. All images were transferred to a commercial medical image processing workstation (uWS-MI, United Imaging Healthcare) for the image evaluation and quantitative analysis.
Time-activity curve measurement
The radioactivity in major organs was obtained from reconstructed dynamic PET to low-dose CT fused images by manual contouring based on anatomical CT images. The volume of interest (VOI) contouring method was based on the study by Deloar H.M. et al. [4] The TACs of 6 organs in each subject, including brain, heart (left ventricle myocardium only), kidneys, liver, lungs, and urinary bladder, were obtained. VOI analysis was carried out with a built-in PET image analysis software uWS-MI. To ensure the accuracy of the activity, a radioactive standard source of 1 ml tube of F-18 FDG solution was used to calibrate the activity. The activity measured both by PET reconstruction and radiation dosimeter was recorded to calibrate the activity measured by PET. In that way, an extra calibration source was placed next to the subject during acquisition to confirm the calibration of the system besides regular quality assurance.
Cumulated activity calculations
Because the decay and most of the metabolic processes are exponential, multi-exponential function is widely adopted to fit TAC data in PET images and we followed the previously reported models for multi-exponential fitting of the short-term data in the present work [2,3,4, 7]. Recently, the time-activity curves represented by a sum of temporal basis functions, for instance, B-splines, have been proposed in parametric estimation in dynamic PET [22,23,24]. For spline method, the area under TAC was calculated as follows:
$$ \overset{\sim }{A_s}={\int}_0^TA(t)\ dt+{\int}_T^{\infty }{A}_f{e}^{-\lambda t} dt $$
(1)
where \( \overset{\sim }{A_s} \) is the cumulated activity, A(t) is the spline fitted activity curve, Af is the activity of the organ at the end of the last PET scan when the emission scan was assumed to begin to decrease only by radiative decay, and λ is the decay constant.
The spline fitting and mathematical integration process was done using Origin (version 8.1).
The time activity curve for urine remained in the urinary bladder; Au(t) was fitted with the sum of two exponentially decaying functions and one constant as follows:
$$ {A}_u(t)={e}^{-\lambda t}\times \left({A}_1{e}^{-{k}_1t}+{A}_2{e}^{-{k}_2t}+C\right)-{\sum}_{i=1}^{\infty}\varepsilon \left(t-{T}_i\right){A}_{ur}(t) $$
(2)
where A1 and A2 are the intercepts, k1, k2 are biological elimination constants, ε(t) is Heaviside function, Ti is the ith urinary voiding time, and Aur(t) is the activity of urine excreted out of the body. Aur(t) was measured by collecting the urine of subjects after the injection, which used a radiation dosimeter (well counter) to measure the activity.
The time-activity curves were fitted as described in previous studies [4], and residence time was derived to compare the spline method with the multi-exponential method.
To fit the equation (2), 12 frames of the organ activity data were selected from 0 to 60 min out of the 8 h measurement: 12 × 120 s as described in previous studies [2, 3, 7], and the data at 10, 40, 70 min: 3 × 180 s were also used for comparison [4].
In our current clinical practice, a PET scan of 20-min duration at 1-h post-injection is used. To evaluate the possibility of individual dose estimation with standard clinical protocols, 5 × 180 s frames of data from 57 to 75 min was used to estimate the cumulated activity, and this was compared with the estimation using the long-time, delayed scanning protocol.
Bladder wall absorbed dose calculation
Previous studies have shown the S value brings the most uncertainty in internal dose calculation of bladder wall absorbed dose among all the organs [16, 17]. In this study, the absorbed dose of bladder wall that was caused by other organs was calculated using S-value obtained from Monte-Carlo simulation. The absorbed dose for the bladder content irradiating the bladder wall was calculated using the dynamic model, which is a function of volume and activity of the bladder content. A dynamic absorbed dose calculation method was based on the assumption that the bladder wall is primarily contributed by two parts of irradiation: the gamma photons and positrons [18, 19, 25, 26]. The average bladder wall dose per unit administered activity was described as,
$$ \frac{D_s}{A_0}=\frac{1}{A_0}\int \left[{D}_{\gamma }(i)+{D}_{\beta }(i)\right] dt=\frac{1}{A_0}\int \left[\frac{\phi_{\gamma }{A}_u(t)}{{\left(36\pi \right)}^{1/3}V{(t)}^{2/3}}+\frac{\Delta _{\beta }{A}_u(t)}{2V(t)}\right] dt $$
(5)
where Ds is the absorbed dose contributed from urine content to the bladder wall, Dγ is the absorbed dose contributed gamma photons, Dβ is the absorbed dose contributed by positrons, ϕγ is the dose conversion factor from the contribution of gamma photons (ϕγ=404.71cm2 ∙ μGy/(MBq ∙ min)), and ∆β is mean positron particle energy emitted per nuclear transition of the radionuclide (∆β=2.3 × 103g ∙ μGy/(MBq ∙ min)) [20].
In this study, we assumed that the urine increases at the same rate between two successive voids:
$$ V(t)=\left\{\begin{array}{c}{V}_0+\int u(t) dt;0\le t<{T}_1\\ {}{V}_b\left({t}_i\right)+\int u(t) dt;{T}_{n-1}\le t<{T}_n\end{array}\right. $$
(6)
where u(t) represents urine production rate.
For dynamic model considering two voiding, absorbed dose after second voiding was assumed to be zero.
Organ absorbed dose calculation
In order to make a comparison of changes of effective dose using different measurement methods, it is necessary to calculate the organ absorbed dose to estimate the effective dose. The calculation of the dose was made via MIRD method as follows [27, 28]:
$$ \frac{D_{rk}}{A_0}=\sum \frac{\overset{\sim }{A_i}}{A_0}S\left({r}_k\leftarrow {r}_h\right)=\sum {\tau}_lS\left({r}_k\leftarrow {r}_h\right) $$
(9)
where Drk is the absorbed dose to a target organ, S(rk ← rh) is the mean absorbed dose to a target region per unit cumulated activity in a source region, rk is the target organ, rh is the source organ, and τ is residence time.
S value calculation
ICRP reference phantom is widely used in absorbed dose estimation [29]. However, the phantom was not based on the typical habitus found in the Chinese population. We revised the ICRP realistic reference phantom into two special phantoms for a man (height, 170 cm; weight, 60 kg) and a woman (height, 160 cm; weight, 51 kg) based on two subjects to carry out the Monte-Carlo simulation.