Preclinical imaging is established as an integral part of biomedical research and the associated development pipeline for new biomarkers, drugs and therapies [1
]. Whilst much effort has been focussed on optimisation of contrast agents, hardware design and image reconstruction, much less attention has been paid to the animal handling and physiological monitoring apparatus. Although many preclinical imaging systems are supplied with the basic apparatus for animal restraint, physiological monitoring and generation of cardio-respiratory gating/triggering signals, most of these apparatuses are not comprehensive and are incompatible across vendors where multi-modal imaging is required [4
Many small-animal imaging and image-guided radiotherapy procedures require the use of general anaesthesia primarily for immobilisation and for minimising stress related to the insertion of invasive devices. Whatever the reason, they all necessitate reliable means to position the animal within the scanner or radiotherapy device, monitor the animal’s vital signs, assess anaesthetic depth and provide thermoregulation in a manner that is compatible with the imaging and/or radiotherapy systems being used. Changes in the cardio-respiratory rate, pattern and amplitude can be monitored, and although some experience is needed to interpret their significance, they can be indicators of anaesthetic-related physiological distress as well as underlying disease. In addition to the animal welfare considerations, cardio-respiratory monitoring also allows for cardio-respiratory gating/triggering to synchronise the imaging to their respective cycles and avoid motion-related image degradation [6
Consequently, there is an opportunity for optimisation and standardisation of the animal handling and monitoring apparatus across scanner modalities and manufacturers. Successful deployment will help improve the quality of physiological maintenance and animal welfare, maximise image quality, increase statistical power and also lower the barriers to imaging and expand the research capabilities of the preclinical imaging techniques [2
]. However, certain requirements will need to be fulfilled to make this optimisation successful and generally applicable. Such multi-modal, multi-vendor compatible animal handling apparatus should be simple and cheap to manufacture, non- or minimally invasive and compact enough to be used in small confines, as dictated by the small imaging fields of, for example, magnetic resonance imaging (MRI) coils, positron emission tomography (PET) detectors or single-photon emission computed tomography (SPECT) collimators.
We have designed and implemented a multi-modal and multi-vendor compatible animal handling apparatus that is used in conjunction with a custom-made physiological signal processing unit; the latter allows cardiac- and respiratory-synchronised imaging to be performed on animals that are moved, in the cradle, between scanners. The animal support cradle can be adapted on a per application basis and features integrated tubing for anaesthetic and tracer delivery, an electrically driven rectal temperature maintenance system and respiratory and cardiac monitoring; all are PET-, SPECT-, computed tomography (CT)- and MRI-compatible and largely insensitive to false activations arising from local environment changes such as doors opening when using a pressure transducer for respiratory monitoring. The associated gating/triggering control signal generation is user-programmable and can be deployed on any system capable of using transistor–transistor logic (TTL) (or similar) gating control signals.
3. Results and Discussion
This cradle system has been used successfully in thousands of scans by several groups over several years. It is currently installed, at our institute, on three MRI scanners (Bruker and Varian), PET/SPECT/CT (MILabs) and an image-guided radiotherapy system (SARRP) upon which MR image-guided radiotherapy is performed in abdominal tumours as a matter of routine [12
]. Further modifications continue to be made to it, and this is made easy by the availability of spare mounting holes on most components and the rapid production of new components on the 3D printer.
A sagittal-view maximum-intensity projection (MIP) image of a mouse in the cradle with a fibre-optic thermometer and heating element, the GPERM and hydrogel ECG electrodes showed no unacceptably high signal intensities derived from the animal support apparatus, indicating its CT compatibility. MRI compatibility has been previously demonstrated, though not described in detail [9
]. A sagittal-view MIP presentation of a CT image is shown in Figure 10
None of the physiological monitoring devices are the brightest objects in the image and so will not present a significant source of image corruption through high levels of X-ray absorption and scattering.
The time required for transfer of the animal from the anaesthetic induction box to the scanner and the start of the first scanner operation is typically <60 s, whilst transfer of animals between scanners is below <90 s. These processes are very efficient, present a very low technical burden to the operator and are robust.
Multi-modal SPECT/CT/DCE-MRI was performed, and images could be easily co-registered, displayed, analysed and presented. The renal pelvis and calyxes, as rendered from the SPECT image following renal excretion of 99mTc-DTPA, overlaid well with the kidneys, as rendered from DCE-MRI. No image distortions attributable to material selection were observed, the animal handling apparatus was easily transferrable between MRI and CT/SPECT and cardio-respiratory signals allowed for prospective gating and triggering for MRI and CT/SPECT, respectively.
shows a tri-modal, cardio-respiratory-synchronised image produced using scanners from two different manufacturers, from the same animal positioned in the same cradle and all acquired within 5 min of each other. The experiment time defined as from induction to full recovery of anaesthesia was 70 min for a single animal. However, throughput can be increased significantly by using two cradles simultaneously and starting the next animal on MRI whilst the previous animal is transferred to SPECT/CT, similar to the workflow described for MRI-guided radiotherapy [12
]. This high throughput is enabled, in significant part, through the use of well-designed animal handling cradles that are both easy to produce in bulk and easy to use. The physiological monitoring is minimally invasive, and excellent-quality ECG and respiration signals result. Though only skin contact is used to detect the ECG, the signals are sufficiently strong and insensitive to the imaging gradients to allow prospective gating control. As a result, the speed advantages described previously [6
] are enabled without the use of invasive needle positioning and whilst avoiding the presence of metal in the CT imaging FOV. The respiration signals are produced through gentle contact of a flexible sheet with the animal, with no requirement to apply pressure to the body as so often required when using pneumatic pressure balloons, and so, there is no need to invoke a postural defect.
The cradles are quick and very simple to assemble; printing of a cradle main tube, hexbase base slider, cable grippers and tooth bar components takes approximately 6 h using our F170 printer. Removal of space-filler material in a solvent bath takes a further 6 h, and post-bath drying takes <3 h. Physical assembly of the prepared components takes <1 h. It is, therefore, quite possible to produce a new cradle from scratch within one working day of demanding it.
The cradle main tube has a flat base with a variable depth and height tooth bar, making it very easy to position the animals with repeatable prone posture. The underside of the flat base also allows the sheet resistor to be mounted very easily and to give a good quality of heating, as described previously [9
]. The upper side of the flat base has a series of ridges spaced by 20 mm that allow a variety of devices, including the ECG plate, to be clipped into a variety of positions along the length of the cradle. Good thermal contact of animal to cradle is achieved, and a range of animal sizes can be accommodated, even when ECG signals are required.
Production of GPERMs is a simple modification of a design described previously [15
] and requires approximately 1 h of printing time, 1 h of solvent bath time and 1 h of drying time. Once all the required parts are available, the assembly of a prepared piezo-electric strip into the housing takes <10 min. Preparation of the piezo-electric strip requires a series of straightforward though intricate and sequential procedures, and so strips are prepared in batches of five or more not individually. A jig, also prepared on the 3D printer, was used to cut the PDVF sheet into 6-mm-wide strips whilst protecting the operator’s fingers from the sharp blade used. Masks for painting the graphene strips were prepared using envelope-label stickers that were cut using a laser cutter, and the laminator sheets were cut into strips again using a jig similar to that used for cutting the PVDF sheet. The thin layer of graphene, painted on a fluorocarbon polymer, is compatible with both MRI and CT (and by inference PET and SPECT) and did not result in image distortions (Figure 10
). The detector is thin, flexible and sensitive, generating respiration signals robustly and without the need to apply pressure to the body with the consequent postural distortion. As described, the respiration monitor provides a rather global measurement of respiration and so may not be optimised for any specific organ that is subject to respiratory motions. A simple redesign of the pad using, for example, a smaller area of painted graphene could be used to optimise the measurement of the breathing motions locally. The design of the entire apparatus allowed easy and quick transfer of the animal handling apparatus between different imaging systems through the use of quick-fit adaptors.
Production of the ECG plates was very fast; each base takes approximately 10 min to print. Solvent stripping and drying takes a further 60 min, but the plates are small enough that they can often be produced in bulk during the print runs for other objects. Once all the required parts are available, the assembly of an ECG plate takes <10 min. The units were formed by shortening the commercial pads’ carbon wires, which were spiralled into a twisted pair in order to minimise pickup of gradient-switching-induced noise in MRI and were attached to quick-fit connectors. The pads were mounted on a support plate, which clips in place onto the ribs of the cradle base, making it easy to adjust its position in relation to the animal. These ECG pads allowed generation of robust ECG signals, even in the presence of high-demand gradient switching during MRI. The sheaths of these wires were relatively bright in CT but no brighter than many of the bones and so do not create any additional confounding source of streak artefact [9
]. In this report, we describe their use in contact with the shaved chest, but they can be expected to work wherever contact with the skin can be achieved.
A set of apparatuses has been described that enables multi-modal imaging to be performed using scanner equipment from multiple vendors, each of which is designed without reference to any other. This allows the advantages of each technique to be exploited, whilst maintaining the ability to match organs on a pixel-wise basis and to provide a multi-parametric characterisation of anatomy, physiology and pathology. This provides clear opportunities for improvements in both scientific content and animal welfare as better data can be produced from fewer animals, especially when paired statistics that can be applied across both the time (repeated imaging) and image modality dimensions are used. In some cases, e.g., SPECT-CT, PET-CT and PET-MRI, multi-modal imaging can be performed within a single scanner assembly, in which case the issues of cradle transfer are largely irrelevant. However, in most cases, separate MRI, PET and SPECT systems would be used, and a SPECT-MRI system does not yet exist. As such, the cradle system described (and any variation on a theme) provides an opportunity to standardise animal handling procedures in a range of environments (scanners) and achieve improved performance, so aiming for maximum reproducibility [16
]. Whilst there has been significant effort in defining how well scanners should operate [17
] and how in vivo experimentation should be practised [18
], there is little detail on how to achieve optimal scanner performance with best practises in animal welfare. The need for immobilisation of the animal through the use of general anaesthesia is recognised [20
], but methods for maintaining temperature and monitoring physiology across a range of multi-modal, multi-vendor imaging systems are not well described. Through a combination of careful material and device selection, we have described an approach that allows animals to be transferred whilst under general anaesthesia between any of the tomographic scanners we currently or have previously operated (Bruker and Varian MRI, MI-Labs PET/SPECT/CT, Xstrahl SARRP radiotherapy and Siemens Inveon PET-CT). Extension to other systems from other suppliers is expected to be straightforward. The heaters, ECG and respiration monitors are easy to use and compatible across all of the cradle sizes we use (23–39 mm diameter). The smallest cradles are designed to allow use with a 26 mm internal diameter, 100 mm long, whole-body RF coil for MRI [8
] and with the high-resolution, small-bore mouse collimators on the MILabs PET/SPECT/CT system.