Optical module for small animal SPECT
The prototype optical imaging (OI) module was fitted onto the U-SPECT-II [21],[22] installed at the LUMC (Leiden, the Netherlands) as is shown in Figure 1. The optical module consists of three main components: 1) a light tight `dark box' ⑤, 2) a very sensitive CCD camera ③, and 3) a bright light source ⑥. Details about these components will be given in later paragraphs. The dark box was designed in such a way that when the module is in `open' position, the handling of the animals in the bed of the U-SPECT is not hampered. When the module is `closed', the CCD camera on top of the box is shielded from ambient light and can produce a total-body top-view bioluminescent image of the animal via a mirror ⑪. For photographic and fluorescence imaging, the animal is illuminated by the light source via two optic fibers ④ entering the box and small mirrors ⑩ reflecting the light onto the bed ①. Excitation and emission light filters, well adapted to the spectral profile of the fluorescent dye under study, can be added to the system.
For SPECT scans, the animal bed is moved by an XYZ stage into the U-SPECT ⑨ collimator. For the OI position, the bed is held in the same position as used for placing an animal in the bed (U-SPECT `bed eject position'). Alternatively, by mounting the OI module on a separate `docking station', it can also be used as an independent optical imaging device. The docking station allows for parallel use of the SPECT and OI module.
Dark box
The dark box uses a rail system to slide over the table, enabling animal placement. The sliding mechanism consists of a U-profile mounted on the box and rails ⑧ mounted on the table that contains the XYZ stage to move the bed mounted at the front side of the U-SPECT. When the box slides to its closed position, it is lowered by about 1 cm. In this way, the U profiles lock over the rails. Blackboard paint has been applied on the rails and the U-profile to ensure the light tightness in locked position.
To further block ambient light, all joints on the box have been sealed with Sikaflex-221 (Sika AG, Baar, Zug, Switzerland) sealant. Blackboard paint is applied to the inside of the box to avoid light reflections. A black ABS-plastic insert ⑦ has been fitted to cover the entrance of the U-SPECT to prevent leakage of ambient light into the dark box and to avoid reflections from the metallic SPECT entrance. This insert was made with a tight fit to the entrance of the existing scanner, leaving maximum space to handle the animals on the bed. With the insert in place, the U-SPECT vision cameras can still be used to select the region of interest (ROI) in the U-SPECT acquisition software [21],[22]. Remaining light leaks measured inside the closed black box were closed using tape sealants.
Camera, lens, and mirror
To allow for the detection of weak bioluminescence and fluorescence signals, an Andor iKon-M 912BV scientific CCD camera (Andor Technology plc., Belfast, UK) with a very low dark current level was used for recording OI images. This camera is based on a CCD77-00 (e2v Technologies Ltd., Essex, UK) with 512 × 512 pixels (24 × 24 μm) and has a quantum efficiency between ~50% and ~95% in the wavelength region of 400-900 nm. With the 16-bit camera read-out at a rate of 50 kHz, the read noise was below 2 counts/pixel. By means of an air-cooled Peltier element, the CCD operating temperature was set to -65°C, reducing the dark current to ~0.006 elec/pix/s. The camera was fully controlled by the accompanying Andor Solis software.
The CCD camera was equipped with a Fujinon (Fujifilm Corp., Tokyo, Japan) CF25HA-1 lens ⑬. This lens has a fixed focal length (F) of 25 mm. To collect as much of the emitted luminescence as possible, we selected a lens whose diaphragm could be set as large as F/1.4. An emission filter holder ⑫ (50 mm diameter) was mounted in front of the lens. A lens hood was integrated with the filter holder to avoid stray light entering the lens/filter combination.
A mirror was placed above the animal bed (Figure 1) to direct the emitted light to the CCD camera and generate a top-view image of an entire mouse. The selected Edmund Optics enhanced aluminum mirror (Edmund Optics Inc., Barrington, NJ, USA) has a high reflective index in the wavelength region of interest: between ~85% and ~98% for wavelengths larger than 450 nm.
Fluorescence excitation light
To excite fluorescent dyes, an illumination setup was constructed to expose the whole animal (bed in OI position) with a relatively high intensity of (excitation) light at the appropriate wavelength.
An MI-150 fiber optic illuminator (Edmund Optics Inc., Barrington, NJ, USA) with a 150-W EKE halogen light bulb produced a high intensity bundle of light (450 to 800 nm), covering our wavelength range of interest for fluorescence imaging. Light from the illuminator was guided to an excitation filter box ② using a 0.25 inch glass fiber bundle guide. In the filter box, the spectrum of the excitation light can be tailored to a particular dye using a 12.5-mm diameter filter. After the filter, the light was split into two fiber bundles that enter the dark box next to the camera. The light from the fibers was reflected by two small mirrors above the bed to illuminate the mouse from two directions. Light leakage at the fiber bundle entrance points was prevented by applying cable glands.
Imaging with U-SPECT-BioFluo
With the U-SPECT-BioFluo device bioluminescence, fluorescence, and SPECT scans of a mouse can be made in a sequential fashion. The SPECT measurement settings are setup as described earlier for the U-SPECT [21],[22]. To perform the optical measurements, the OI module is closed and the animal bed is moved to the OI position. After positioning, grayscale photographs and bioluminescent and/or fluorescent images can be recorded by the CCD camera and stored with the Solis software for further offline analysis with Matlab (MathWorks Inc., Natick, MA, USA).
Grayscale photograph
Grayscale images are recorded by the CCD camera to provide an anatomical context for the bioluminescence, fluorescence, and SPECT images. The grayscale images are recorded under white light illumination without the use of emission and excitation filters. To avoid saturation of the CCD image for our minimal exposure time (50 ms), the lens diaphragm is set to F/22 and the iris of the light source is set to a minimum. The F/22 diaphragm allows for sharp photographs of the animal in OI position.
Bioluminescence imaging
The bioluminescence images are recorded without the use of excitation light and an emission filter. To provide total darkness in the module, the fiber optic illuminator is switched off and a light stop is put in the excitation filter box to block light entering the module via leaks in the illuminator. To collect a maximum of bioluminescent photons in a reasonable exposure time (<90 s), the largest possible lens diaphragm (F/1.4) is used.
Fluorescence imaging
During fluorescence imaging, the object or animal is continuously illuminated using an appropriate excitation filter for the dye under study. With a specific emission filter in front of the CCD camera lens, only the emitted fluorescent signal is recorded by the camera.
The CCD exposure time was kept to a minimum by using the largest possible lens diaphragm (F/1.4) and fully opening the iris of the illuminator. During imaging, the exposure time was adjusted to obtain images in which the brightest fluorescence spots have pixel values larger than half the full pixel ADC range.
Illumination calibration
The measured images are corrected for the non-uniform illumination pattern at the mouse bed by making calibration measurements. This is done by making an image (without emission filter) of the illumination pattern using a flat sheet of white paper in the OI position. The corresponding iris, diaphragm, and excitation filter are used for the grayscale or fluorescence calibration. The exposure time has to be adjusted to avoid CCD saturation. Since the bioluminescent images do not require illumination, they do not require such a calibration.
Image processing
The optical images from the CCD camera are processed offline with Matlab. Both the fluorescence and photographic calibration images are blurred with a Gaussian filter of 7.30 mm full width at half maximum (FWHM) to suppress small scale non-uniformities of the calibration paper sheet itself. The resulting calibration images are subsequently normalized by scaling their maximum pixel value to 1. The photographic and fluorescent images are corrected for the non-uniform illumination by dividing them by the corresponding normalized calibration image.
3D SPECT images are reconstructed from the SPECT list-mode data with MILabs reconstruction software version 2.38. Proper energy and background windows are set and data is reconstructed to an isotropic voxel grid of 0.2 mm using a pixel-based ordered subset expectation maximization (POSEM) algorithm [25] which includes compensation for distance-depending blurring [26]. No attenuation correction is applied. For comparison to the optical images, a (2D) SPECT sum image is made by summing the voxel values along the vertical direction. For anatomical reference, the bioluminescence, fluorescence, and SPECT sum images are shown in color on top of the grayscale photographic image for pixel values above an adjustable threshold. For bioluminescence and fluorescence, this is straightforward as the same CCD camera is used to record these images. Overlay of the SPECT sum image was done by matching the mouse contours to the grayscale images of the mouse.
Performance characterization
To evaluate the multimodal imaging capabilities of the U-SPECT-BioFluo setup, pilot experiments were performed using a phantom and tumor-bearing mice injected with a multimodal tracer. Optical images were also acquired and processed on an IVIS Spectrum with Living Image software (Caliper Life Sciences Inc., Hopkinton, MA, USA). Bioluminescence imaging on the IVIS was done with open filter settings.
The fluorescent dye CyAL-5.5b (λex,max =674 nm; λem,max =693 nm) [27] was used for the phantom and the multimodal tracer. For the fluorescence measurements on the IVIS, the (built-in) 640-nm excitation and 680-nm emission filters were selected. Their bandwidths were respectively 30 nm and 20 nm. However, for the OI module we selected a commercial filter (Edmund Optics TechSpec Fluorescence Bandpass, Barrington, NJ, USA) with a center wavelength of 692 nm and wider bandwidth of ~40 nm to match the dye's emission peak and to detect a larger part of the emission spectrum. From the same filter series another filter with ~40 nm bandwidth was used to excite this dye. To minimize the spectral overlap of the excitation and emission filters, an excitation center wavelength of 624 nm was selected. The excitation light spectrum of the filter was largely overlapping the dye's absorption spectrum. Both commercial filters have a 93% transmission within their passbands and a blocking factor >106 for wavelengths outside their passbands. This blocking factor is lower for non-perpendicular incident light. So, some excitation light reflected by the object and incident on the emission filter might still pass that filter causing the so-called excitation light leakage background in fluorescence images,
Phantom measurements
A phantom was generated, via a slightly modified procedure as described by Pleijhuis et al. [28], to evaluate the attenuation and resolution for the fluorescence and SPECT signals as function of tissue depth. This phantom (Figure 2) consisted of a cylinder of tissue simulating gel, in which a capillary, uniformly filled with a fluorescence/SPECT compound, was inserted under an oblique angle.
The gel consisted of a 1% (w/w) agarose solution that was prepared by dissolving agarose (Roche, Basel, Switzerland) in 50 mL water under continuous heating and stirring. After cooling down to 40°C, 170 μM hemoglobin from bovine blood (Sigma Aldrich, St. Louis, MO, USA) was added to the solution to obtain an optically tissue equivalent gel [28]. The solution was poured into a 50-mL tube (Falcon, BD Biosciences, San Jose, CA, USA). After further cooling and solidification, the tube was removed and the gel was given blunt ends using a scalpel.
A 200-μM CyAL-5.5b solution was mixed 3:1 with a 99mTc-eluate (~1.3 GBq/mL). A glass capillary (inner diameter 0.9 mm) containing 50 μL of this mixture and closed by epoxy glue [29] was inserted into the gel. Prior to inserting the capillary in the phantom, we used the blue coloration to visually confirm that the fluorescent dye (CyAL-5.5b) was uniformly distributed in the capillary.
The phantom was placed in the U-SPECT-BioFluo bed, and a grayscale and a fluorescence image were recorded. Subsequently, a 20-min SPECT scan was made with the ROI containing the entire phantom. At last, fluorescence measurements were made on the IVIS for control.
In vivo measurements
Mouse scans were performed to interpret the different bioluminescent, fluorescent, and SPECT images for a practical preclinical case series of four animals. Tumor lesions were generated in Balb/c nude mice (6 to 8 weeks of age) by transplanting 4 T1-luc + tumor cells (0.25 • 105), with firefly luciferase gene expression, into the fourth mammary fatpad [30].
For fluorescence and SPECT imaging, the previously reported multimodal tracer 111In-MSAP-RGD [30], containing the above mentioned CyAL-5.5b dye, was used. Twenty-four hours prior to imaging, the mice were injected intravenously with this tracer (40 μg, 18 nmol, 10 MBq).
For bioluminescence imaging, the mice were injected intraperitoneally with 150 mg/kg D-Luciferin (Xenogen Corp., Alameda, CA, USA) in PBS. Ten to twenty minutes after this injection, the mice were placed in the transparent U-SPECT-BioFluo bed and first imaged on the IVIS. Afterwards, the bed with the mouse was immediately clicked on the stage of the U-SPECT-BioFluo for optical imaging followed by 30 min of SPECT imaging. The mice were anesthetized by means of Hypnorm/Dormicum/water (1:1:2; 5 μL/g i.p.). The protocol timeline is shown in Figure 3. These animal experiments were performed in accordance with Dutch welfare regulations and approved by the ethics committee of the Leiden University Medical Center under reference 2013/12189.