Introduction

Molecular imaging of small animals with single photon emission tomography (SPECT) and positron emission tomography (PET) has recently emerged as an important tool with which to assess molecular mechanisms and pharmaceutical development [1]. Sub-half-millimeter (mm) SPECT resolution is useful for systems equipped with multi-pinhole collimators [2, 3], but sub-mm resolution has recently been achieved for PET tracers using pinholes [4]. SPECT and PET play complementary roles in multimodality imaging [5, 6]. Thus, the simultaneous SPECT and PET imaging might gain new insight into the multiple biologic mechanisms by visualizing and quantifying different aspects of biologic function.

Multimodality preclinical imaging systems have recently been integrated into a common gantry and are now commercially available [7]. Examples of such preclinical trimodality systems are the versatile emission computed tomography/computed tomography (VECTor/CT) [8, 9], Triumph™ Trimodality System [10], Inveon PET/SPECT/CT [5] and ALBIRA [11, 12]. Although the Triumph, Inveon and ALBIRA systems cannot simultaneously acquire SPECT and PET images, VECTor enables fully simultaneous SPECT/PET image acquisition based on clustered-pinhole collimation [8].

Pinhole edge penetration by 511 keV annihilation photons renders current multi-pinhole collimators unsuitable for the high-resolution imaging of positron emitters [4]. Although reducing pinhole acceptance angles can reduce resolution loss due to edge penetration, such improvement is achieved at the cost of each pinhole having a reduced field of view (FOV). Goorden et al. [4, 13] developed a novel collimation system using clustered pinholes to simultaneously and separately image γ-photons emitted by PET and SPECT tracers. The central FOV in this clustered multi-pinhole system is kept equal to that of the conventional pinhole system by selecting an opening angle of 16° and 18° for each pinhole in the cluster [8].

The manufacturers have recently published information about the basic performance of the VECTor system [8]. Although they have proven that the image quality in terms of resolution, contrast, and the count-to-noise ratio is superior to conventional PET system [9], the detailed performance under several different conditions has not been evaluated yet. The present study experimentally evaluates the performance of the clustered multi-pinhole system for simultaneous 99mTc and 18F imaging from the user perspective.

Materials and methods

System and collimator geometry

The VECTor system (MILabs, Utrecht, Netherlands) designed for simultaneous SPECT and PET imaging of small animals. This system comprises three stationary gamma detectors in a triangular configuration. The NaI(Tl) crystals have an active area of 51 × 38 mm2 and a thickness of 9.5 mm. The exchangeable cylindrical clustered-pinhole collimators are placed in the center of the system. The clustered-pinhole collimator designed for high-energy photons (511 keV) is previously described [4]. The clustered-pinhole collimator made of tungsten contains 48 clusters of four 0.7-mm-diameter pinholes placed in four rings (Fig. 1a, b). Of 192 clustered pinholes, 162 were used for image reconstruction. The collimator has a wall thickness of 43 mm, an inner diameter of 48 mm and the pinhole centers are placed at a diameter of 64 mm. The diameter of the central FOV in the clustered-pinhole collimator described herein was 12 mm and the longitudinal length was 9 mm.

Fig. 1
figure 1

Clustered-pinhole collimator optimized for imaging SPECT and PET tracers. a The collimator of tungsten comprising four rings. b Clusters of four 0.7-mm-diameter pinholes (white circle). The collimator contains 48 clusters placed in four rings (inner view)

Full size image

Data acquisition and image reconstruction

Data were acquired using the scanning focus method as proposed by Vastenhouw and Beekman [14]. The photopeak windows of all SPECT and PET data in this study were set to a width of 20 %. Images were reconstructed using a pixel-based ordered-subsets expectation maximization (POSEM) algorithm with 16 subsets and 10 iterations. All SPECT and PET images were reconstructed on a 0.4 mm voxel grid with MILabs reconstruction software (β version).

Scatter correction used a triple-energy-window (TEW) technique [15] and was performed on the all SPECT and PET data. The background windows were set to 100–115 and 155–170 keV for 99mTc tracers and 400–450 and 550–600 keV for 18F tracers. No attenuation correction was applied.

Evaluation of system characteristics

Spatial resolution and sensitivity of stand-alone 99mTc and 18F imaging

Two hot-rod micro-spatial resolution phantoms with 6 sectors were created to test the resolution. The capillary diameters were 0.35, 0.40, 0.45, 0.50, 0.60 and 0.75 mm, in the phantom containing 99mTc and 0.45, 0.5, 0.55, 0.75, 0.8 and 0.85 mm in the phantom containing 18F. The capillaries contained 40 MBq of 99mTc and 44 MBq of 18F at the start of scanning.

Sensitivity was measured by scanning 99mTc and 18F point sources with 15 MBq activity positioned at the center of the FOV. The acquisition time was 10 min. Sensitivity was calculated by dividing the counts in the photopeak by the measured activity using Eq. 1:

$${\text{Sensitivity}}\left( {{\raise0.7ex\hbox{${\text{cps}}$} \!\mathord{\left/ {\vphantom {{\text{cps}} {\text{MBq}}}}\right.\kern-0pt} \!\lower0.7ex\hbox{${\text{MBq}}$}}} \right) = \frac{{{\text{Count}} \text{-} {\text{rate}}\,\,{\text{in}}\,\,{\text{photopeak}}}}{{A_{t0} }}$$
(1)

where A t0 represents the measured activity.

Count rate linearity

Count rate as a function of activity was measured using a point-source phantom comprising a 0.2-mL micro-tube. A phantom with a volume 0.1 mL containing either <800 MBq of 99mTc or 418 MBq of 18F was positioned along the axis of the scanner and centered in the FOV. The phantom was fully covered by the 3-dimensional central FOV (an ellipsoid of 12 × 12 × 9 mm).

Nine phantoms contained 3, 6, 12, 25, 50, 100, 200, 400 and 800 MBq of a homogeneous 99mTc solution. The photopeak counts at each time frame with a 30-min acquisition period were calculated to determine count rates.

We used the decay method to generate count response curves for an initially high 18F activity concentration that was allowed to decay. List-mode acquisition proceeded for 20 h (11 half-lives of 18F) until the activity decayed to <0.2 MBq. Each 30-min section of list-mode data was binned and the photopeak counts were recorded.

The count rate linearity performances of 99mTc and 18F were evaluated by the relationship between the measured count rates and the activity.

Spill-over of 18F into 99mTc SPECT images

We quantified the influence of 18F on 99mTc SPECT images as follows. One point-source phantom with a volume of 0.1 mL was filled with a mixture of 99mTc (37 MBq) and 18F (370 MBq). The other was filled with only the 99mTc (37 MBq). The phantom was fully covered by the central FOV. List-mode acquisition proceeded for 12 h/scan. The SPECT images were reconstructed from the list-mode data. The region of interest (ROI) covering the entire phantom in the projection was used for SPECT. The photopeak counts at each time frame with a 30-min acquisition period were calculated. Ratios of SPECT counts on the 99mTc-only and simultaneous 18F-99mTc at various time points (t = 0, 1 and 2–12 h) were evaluated as a function of the 18F-to-99mTc radioactivity concentration ratio.

Results

Figure 2 shows representative images of the spatial resolution phantom containing 99mTc and 18F tracers. The 0.5-mm hot-rods can be visually distinguished in the 99mTc image, and 0.8-mm rods for 18F are visible. The sensitivity measured with 99mTc and 18F was 2800 cps/MBq (0.28 %) for 99mTc and 2899 cps/MBq (0.29 %) for 18F, respectively.

Fig. 2
figure 2

Reconstructed images of 99mTc and 18F in thin hot-rod phantoms. Slice thickness is 0.4 mm

Full size image

Figure 3 shows count rates as a function of activity of 99mTc and 18F. The peak count rates of 99mTc and 18F were 588 kcps at 427 MBq and 264 kcps at 196 MBq, respectively. The slope of the resulting linearity curve for 99mTc was roughly equal to that for 18F at low activity up to 50 MBq, whereas at high activity levels the count rate of 18F was limited.

Fig. 3
figure 3

Count rates as a function of activity of 99mTc and 18F using 0.1-mL point source. Circle 99mTc, square 18F

Full size image

Figure 4 shows the energy spectra of the 99mTc, 18F and simultaneous 99mTc and 18F images. The 99mTc-imaging spectrum showed only the 141-keV photopeak for 99mTc (Fig. 4a), whereas 18F-imaging spectrum showed that energy spectra except for the primary photopeak were measured at 170 keV in the photopeak windows (Fig. 4b). In simultaneous 99mTc and 18F images, three events at 141, 170 and 511 keV were measured (Fig. 4c). The energy spectrum of 99mTc was slightly distorted due to the close proximity of the photopeaks of 141 and 170 keV.

Fig. 4
figure 4

Spectra with clustered-pinhole collimator in stand-alone 99mTc (a) and 18F (b) images and in simultaneous 99mTc and 18F image (c). The activities for 99mTc and 18F were about 37 and 370 MBq, respectively. Black arrow photopeak of 170 keV originated from 18F at high photopeak energy

Full size image

Figure 5 shows SPECT count ratios of 99mTc-alone and 18F-99mTc simultaneous SPECT in relation to the 18F-to-99mTc radioactivity concentration ratio. The SPECT count ratio of 99mTc in stand-alone to 99mTc in simultaneous images remained constant (almost 1) up to an 18F:99mTc concentration ratio of 2:1. It exponentially increased with an increase of 18F-to-99mTc radioactivity ratio.

Fig. 5
figure 5

The SPECT count ratio of 99mTc for 99mTc-alone and 18F-99mTc simultaneous acquisition in relation to 18F-to-99mTc radioactivity concentration ratio. For the data, TEW scatter correction was employed

Full size image

Discussion

We evaluated the performance characteristics and imaging capability of a contemporary high-tech SPECT/PET/CT system for imaging small animals. The manufacturers have recently published information about the basic performance of the VECTor system [8, 9]. However, the present study assessed detailed performance characteristics that lead to determination of the optimal administered doses of 99mTc and 18F tracers. The VECTor system achieved high resolution and high sensitivity as well as good 99mTc and 18F linearity.

Visual assessment using a spatial resolution phantom instead of the full width at half maximum (FWHM) of a line phantom is likely to provide spatial resolution estimates closer to the actual values obtained during small-animal imaging studies [16]. In our scanner, an imaging study using two types of phantoms resolved 0.5- and 0.8-mm rods for 99mTc and 18F tracers, respectively, showing the actual sub-mm imaging capability of the VECTor system. Previous reports have indicated that spatial resolution for 99mTc SPECT of Triumph [10], Inveon [5] and ALBIRA [11, 12] are 0.4, 1 and 0.8 mm, respectively, and 1.35, 1.4 and 1.55 mm for 18F PET, respectively. Thus, our results of VECTor were better than that of previously developed devices [11, 17]. Goorden et al. [4] showed that clustered pinholes could offer improved spatial resolution. In addition, our results of stand-alone imaging corresponded to those of previous studies of simultaneous imaging [8]. We consider that the quality of simultaneous 99mTc and 18F images acquired using the VECTor can compete with that of individual 99mTc and 18F images. In general, SPECT has better spatial resolution, whereas PET is more sensitive when imaging small animals. The sensitivity was relatively lower for pinhole, than coincidence PET with Triumph, Inveon and ALBIRA (peak sensitivity: 0.29 vs. 6 %, 6.7 and 2 %, respectively). Nevertheless, we acquired high-resolution PET images that compared favorably with the Triumph, Inveon and ALBIRA PET systems. In fact, most dedicated small-animal SPECT/PET systems use pinhole collimation for the most optimal resolution–sensitivity trade-off [18, 19].

A small-animal SPECT/PET system should be able to handle a wide range of count rates without serious losses in count rate linearity. The count rate performance of 99mTc and 18F tracers with clustered-pinhole collimators has not been evaluated until now. Count rate curves might considerably vary depending on the phantom [20, 21]. We adopted a 0.1-mL point source, which be fully covered over one bed position and the size is similar to that of a mouse organ (12-mm diameter). The slope of the resulting linearity curve for 99mTc was essentially equal to that of 18F, as count rates for both were linear up to an activity of 50 MBq (~120,000 cps), as seen in Fig. 3. These results indicate that the quantitative performance of the VECTor system at activity levels used in most biologic SPECT or PET experiments is excellent. Harteveld et al. [22] have shown that count loss in terms of U-SPECT-II (VECTor without the clustered-pinhole collimator) performance is not so important in view of the highly linear relationship between count rate and source activity. However, resolution might deteriorate at high-count rates due to event pile-up [23, 24]. The VECTor system could handle events in 99mTc and 18F solutions with maximum count rates of 600,000 and 250,000 cps, respectively. Systematic difference in count rate curves could be explained by count loss due to crystal thickness. The VECTor has γ-detectors [NaI(Tl) crystal thickness, 9.5 mm] that are optimized for imaging SPECT tracers. NaI(Tl) crystal is not ideal for positron imaging due to its inherent low efficiency for stopping photons. The NaI(Tl) crystal of the VECTor is 9.5 mm thick, which means that only about 10 % of the 511-keV γ-photons that reach the detector are detected in the photopeak [13].

Clustered pinholes with smaller acceptance angles than traditional pinholes will normally reduce photon penetration [4, 13]. Goorden et al. [13] performed Monte Carlo simulations to quantitatively evaluate the effect of scattered and penetrating photons, and thus indicated that photon scatter from clustered-pinhole collimators was negligible. However, when 18F imaging, energy spectra except for the primary photopeak were measured at 170 keV in the photopeak windows. We consider that the spectrum is mainly due to the backscatter of photons interacting with the clustered-pinhole collimator of tungsten. This is confirmed by the existence of Compton edge of approximately 340 keV (Fig. 4) [25]. In simultaneous 99mTc and 18F imaging, the image reconstructed with the counts of primary photons emitted from 99mTc with low photopeak energy is distorted by backscatter photons that originated from 18F at high photopeak energy. Thus, such photons might influence quantitative and imaging data when dealing with high-activity 18F.

Activities are typically higher in SPECT than in PET studies [26]. However, studies using VECTor require higher doses of PET tracer because the pinhole PET has only one-tenth of the sensitivity of coincidence PET. Thus, some PET tracer might induce spill-over into SPECT images. The limits of equipment performance that can compensate for spill-over must be identified. The ratio of stand-alone to simultaneous SPECT imaging is theoretically constant. However, spill-over from 18F into the 99mTc images would affect this ratio, resulting in an increased ratio of stand-alone to simultaneous SPECT imaging counts when the 18F-to-99mTc concentration ratio is >2. Correction for increasing down scatter from 18F with increasing activity ratio can potentially adversely affect the noise and resolution in the 99mTc image. It has been shown that for ratios <2.9 [8], this effect is very minor and can be tolerated. In this study, we adopted the activities of 37 MBq for 99mTc that typically used in most biologic SPECT experiments [8]. Since the count losses of 99mTc and 18F induced by dead time losses are different as shown in Fig. 3, the starting 99mTc activity might have influenced the results of the acceptable the 18F-to-99mTc concentration ratio. Further studies of spill-over with various combined sources are required.

The presence of large backscatter photopeak (170 keV gamma photons) causes errors in scatter correction based on the TEW method. Although the TEW method provides a reasonable correction for scatter [27], the similarity between the 141 keV primary and 170 keV backscatter of 18F energy spectra causes difficulties with compensating for scatter events using techniques that are based on spectral analysis [28]. We do not consider it enough to use only the conventional TEW method for the scatter correction including backscatter of VECTor. Further study is warranted to elucidate the compensation for 170 keV backscatter on simultaneous 99mTc/18F experiments.

The present study is limited by the fact that all images acquired using 99mTc and 18F solutions were reconstructed under identical conditions. Reconstruction parameters should be carefully selected according to the experimental objectives and tracers because image quality considerably varies depending on the selected values [29]. This trade-off between the recovery of small structures and image noise influenced by the number of iterations and spatial filtering should be considered in the reconstruction setting [30].

Conclusions

We evaluated the performance characteristics and imaging capability of the VECTor system from the user perspective and found that the resolution and sensitivity of 99mTc and 18F imaging is high. Count rates up to 120,000 cps for one bed position were linear for activity, indicating excellent quantitative performance. Spill-over from 18F into 99mTc SPECT images was negligible when the activity concentration of the administered 18F solution exceeded that of the 99mTc solution by up to a factor of 2. Simultaneous 99mTc SPECT/18F PET imaging with two nuclides as well as stand-alone 99mTc and 18F imaging using clustered-pinhole collimators is feasible and practical for a wide range of research applications using small animals.