Design and development of the rat dual support system

The dual support system was developed for rats and dedicated to Siemens INVEON PET/CT scanner. An open-source computer-aided modeling software (www.blender.org) was employed to generate the 3D files for the design of the dual support system (Fig. 1A), created via a 3D-printer (Raise 3D N2 Plus).

Fig. 1 

Design of the dual support system. A Rendering of 3D-printable support (in black) allowing the superposition of INVEON’s original beds. B View of two phantoms within the INVEON dual support bed: in pale green, the designed cone masks and ear bars. C View of two installed rats inhaling anesthetic gas. D Rats monitoring during PET acquisition

This 3D-printed support system is made of PolyLite™ PLA and composed of two clamping holders allowing to fix and overlap one of the original Siemens INVEON beds 5 cm above the main one, attached to the moving table (Fig. 1B). Both beds are equipped with individual cone masks for anesthesia delivery during PET/CT acquisitions (Fig. 1C). To stabilize the rat’s head and so to avoid any movement, stereotactic ear bars as well as tooth bar were also designed and 3D-printed. In the PET scan position, the rear of the animals remains accessible to the experimenter to administer the radiotracer or any other pharmacological agent, including during the scan (Fig. 1D).

In vitro PET study

The dual support system was first evaluated in vitro to quantify the accuracy of the radioactivity measurements for various positions in the scanner (upper, lower or dual). For each experiment, two cylindrical phantoms (Falcon tubes) were filled with 50 ml [18F]FDG solution with activity differing from about 40%. The mean “low dose” phantom activity was 5.8 ± 0.9 MBq, and the “high dose” phantom activity was 8.5 ± 1.2 MBq.

An example of the timeline of one experiment is shown in Fig. 2A. The scanning conditions were: (1) Duo Test: Phantom #1 (higher activity) in the lower bed and Phantom #2 (lower activity) in the upper bed, (2) Phantom #1 alone, in the lower bed, (3) Phantom #1 alone, in the upper bed, (4) Phantom #2 alone, in the upper bed, (5) Phantom #2 alone, in the lower bed, (6) Phantom #1 in the upper bed and Phantom #2 in the lower bed. Each scan has 10 min duration. The scanning procedure details are given in the in vivo section.

Fig. 2

PET experiment timelines: A In vitro experiment: a total of 6 sequential imaging acquisitions were performed during the session. B In vivo experiment: a total of 6 sequential imaging acquisitions were performed during the session

Six experiments were performed. The mean phantom activity ratio was 40% (range 32–52%).

Six images were acquired per experiment, leading to a total of 36 measurements submitted to quantitative analysis.

In vivo PET study

The dual support system was then evaluated in vivo over four experiments of PET scans with [18F]FDG including two rats. All the experiments were conducted in strict accordance with the European Community Council Directive of September 22, 2010 (2010/63/UE) and received approval from the French national ethics committee.

Animal preparation

Eight Sprague–Dawley adult male rats (Charles River laboratories, France) of 330 ± 66 g were used. Animals were housed in standard temperature and humidity conditions with a 12 h/12 h light/dark cycle. Food and water were provided ad libitum. Four hours prior to the experiments, each couple of rats, were food deprived, but they had free access to water. This restriction was applied to selectively maximize and homogenize the [18F]FDG uptake at the brain level, as previously emphasized by Fueger et al. [8]. The animals were anesthetized with 4–5% isoflurane for 5 min (induction phase). A catheter was placed into their caudal vein for radiotracer injection purposes.

The couple of rats were positioned in the dual support system in prone position, the rat #1 underneath the rat #2. During the PET/CT acquisitions, the level of anesthesia was maintained at 2% of isoflurane, with 0.8 L/min air flow rate delivered in a cone mask adapted to rat anatomy (see Fig. 1C). For each rat, the peripheral capillary oxygen saturation level (SpO2) was monitored using the Nonin-9847 V veterinary pulse oximetry sensor system (www.nonin.com, Fig. 1D). To constantly maintain the body temperature at a physiological level, both rats were wrapped in a homemade warming sleeve.

Course of an experiment

After positioning, at T0, the rats received simultaneously a [18F]FDG activity of 34.5 ± 5.5 kBq/g, delivered intravenously. Then, each experiment followed the timeline showed in Fig. 2B including six pet scans in the following conditions: (1) Duo Test: both rats scanned simultaneously (2) Rat #1 Solo Test, in the lower bed while rat #2 is removed, (3) Duo Retest: both rats scanned simultaneously, (4) Rat #2 Solo Test, in the upper bed while rat #1 is removed, (5) Rat #1 Solo Retest, in the lower bed, (6) Rat #2 Solo Retest, in the upper bed.

Forty-eight images were analyzed (6 images/experiments, 4 experiments for a total of 8 animals).

PET/CT imaging procedures

PET imaging was performed using a dedicated small animal PET/CT INVEON system manufactured by Siemens (Erlangen, Germany). The camera has an axial FOV of 12.7 cm and a spatial resolution of 1.8 mm full width at half maximum (FWHM) (in accordance with Bao et al. [7]). Each PET acquisition consisted of a 10 min list mode emission acquisition, followed by a 10 min CT scan using the magnification low acquisition settings for each condition, as represented in Fig. 2. The CT acquisition is used to correct for tissue attenuation and scatter corrections. The acquisition parameters were: attenuation mode; projection: 120; rotation: 200°; binning 4 × 4; effective voxel size: 0.111 × 0.111 × 0.111 mm3; voltage: 80 kV; current: 500 µA; filter thickness: 0.5 mm; exposure: 300 ms. PET acquisitions were reconstructed with attenuation and scatter correction by 3D ordinary Poisson ordered subsets expectation maximization (OP-OSEM3D) with 4 iterations and a zoom factor of 1. The reconstructed image is a volume of 159 slices of 128 × 128 matrix voxels, with voxel size 0.4 × 0.4 × 0.8 mm3. CT data were reconstructed using a Feldkamp algorithm with a down sample of 2 leading to a reconstructed voxel size of 0.2 × 0.2 × 0.2 mm3.

The first PET acquisition (Duo Test) started at T20 timepoint (20 min after [18F]FDG injection), followed by the corresponding 10 min CT scan. Then, the Rat #2 was removed from its bed and isolated outside the PET/CT imaging system on a heating pad and maintained under anesthesia. The second PET/CT acquisition of the day was then performed on the Rat #1 alone (Rat #1 Solo Test) at T45 timepoint (45 min after [18F]FDG injection). At the end of this acquisition, the Rat #2 was placed back in the imaging system to acquire the third PET/CT scan (Duo Retest), 1h10 post [18F]FDG injection. Then, the Rat #1 was removed from its bed and isolated outside the PET/CT imaging system on a heating pad and maintained under anesthesia. The fourth PET/CT scan (Rat #2 Solo Test) was then performed on the Rat #2 alone, 1h35 post [18F]FDG injection. Next, the Rat #2 was again removed, and the Rat #1was placed back for the fifth PET/CT acquisition (Rat #1 Solo Retest), 2 h post [18F]FDG injection. Lastly, the Rat #1 was removed and returned to its cage. The last acquisition of the afternoon was performed on the Rat #2 alone (Rat# 2 Solo Retest) 2h25 post [18F]FDG injection.

Image post-processing and quantification

Image processing was carried out using the Inveon Research Workplace (IRW 4.2) software (Siemens Medical Solutions USA).

In vitro study

A cylindrical volume of interest (VOI) of 6 cm3 was defined and used to measure the radioactivity concentration in phantoms. With in vitro study, we did not attend to reproduce the NEMA protocol, but rather to use a phantom of size and activity comparable to those used in actual rat studies, and to test the different positions of our device. On these phantoms, an arbitrary VOI of a size that could include a rat brain was arbitrary defined. For all acquisitions, the VOI was placed in the center of the phantom volume, to be equidistant from each edge and to avoid overlapping the air bubbles sometimes residing in the phantom (Fig. 3A). The quantified measurements were expressed in Bq/mL and corrected for decay at the start time of the first imaging acquisition.

Fig. 3

ROI analysis. A In vitro experiment showing the cylindrical VOI in the center of the two phantoms. B In vivo experiment, zoom on the brain showing the coregistered ROI atlas on the PET images

Homogeneity of the VOI is defined as the coefficient of variation (i.e., standard deviation/mean) of the voxels activities within the VOI, expressed in percent.

In vivo study

Individual PET images were spatially coregistered over the Lancelot rat brain atlas (Lancelot et al. [9]) to allow automatic delineation of 31 brain regions of the atlas (Fig. 3B). The quantified measurements in brain regions were expressed in Standardized Uptake Value:

$$} = }\;(}/})/}\;(})/}\;\left( } \right)$$

where Activity in the mean activity in the region, corrected for decay at the injection time.

Profiles lines of the brain activity were drawn using IRW (Fig. 8B).

Reproducibility of measurements over the various configurations of scanning is first estimated by the linear regression of brain regional activities (31 regions), pooling the subjects [8], taken the SOLO TEST or DUO TEST scan as the reference.

In the absence of knowledge of the ground truth of the brain activity, the accuracy of activity measurements was assessed in term of reproducibility indexes between scans—bias, variability and the intraclass correlation coefficient (ICC)—defined as follow:

Bias The test–retest bias was calculated as the difference between the test and retest SUV divided by the mean of the test and retest values.

Variability is defined as the standard deviation (SD) of the bias over subjects. These parameters were expressed as percentage units.

Reliability. The measurements' reliability was assessed by ICC calculated as the ratio (BSMSS − WSMSS)/(BSMSS + WSMSS) where BSMSS is the mean sum of square between subjects, and WSMSS is the mean sum of square within subjects. This statistical ratio estimates the relative contributions of between- and within-subject variability and assumes values from − 1 (i.e., BSMSS = 0) to 1 (identity between test and retest, i.e., WSMSS = 0).

A bias under 10%, a variability of 5% and an ICC over 0.7 are deemed to be acceptable.

Reproducibility indexes were computed at the regional level, from the extracted values of each individual region of the 31 ROIs of Lancelot’s rat brain atlas for all the animals (n = 8) across all conditions (Solo test, Solo retest, Duo test, Duo test, Duo retest).