High-Resolution Cardiac Positron Emission Tomography/Computed Tomography for Small Animals

Summary

Here, we present an experimental imaging protocol for the quantification of cardiac function and morphology using high-resolution positron emission tomography/computed tomography for small animals. Both mice and rats are considered, discussing the different requirements of computed tomography contrast agents for the two species.

Abstract

Positron emission tomography (PET) and computed tomography (CT) are among the most employed diagnostic imaging techniques, and both serve in understanding cardiac function and metabolism. In preclinical research, dedicated scanners with high sensitivity and high spatio-temporal resolution are employed, designed to cope with the demanding technological requirements posed by the small heart size and very high heart rates of mice and rats. In this paper, a bimodal cardiac PET/CT imaging protocol for experimental mouse and/or rat models of cardiac diseases is described, from animal preparation and image acquisition and reconstruction to image processing and visualization.

In particular, the ¹⁸F-labeled fluorodeoxyglucose ([¹⁸F]FDG)-PET scan allows for the measurement and visualization of glucose metabolism in the different segments of the left ventricle (LV). Polar maps are convenient tools to display this information. The CT part consists of a time-resolved 3D reconstruction of the entire heart (4D-CT) using retrospective gating without electrocardiography (ECG) leads, allowing the morphofunctional evaluation of the LV and the subsequent quantification of the most important cardiac function parameters, such as ejection fraction (EF) and stroke volume (SV). Using an integrated PET/CT scanner, this protocol can be executed within the same anesthesia induction without the need to reposition the animal between different scanners. Hence, PET/CT can be seen as a comprehensive tool for the morphofunctional and metabolic evaluation of the heart in several small animal models of cardiac diseases.

Introduction

Small animal models are extremely important for the advancement of the understanding of cardiovascular diseases^1,2. Non-invasive, diagnostic imaging tools have revolutionized the way we look at cardiac function in the last decades, both in clinical and preclinical settings. As far as small animal models of cardiac diseases are concerned, specific imaging tools have been developed with very high spatiotemporal resolution. Thus, such instruments can match the need for accurate quantification of the relevant metabolic and kinetic myocardial parameters on the very small and very fast-moving hearts of mice and rats in specific disease models, such as heart failure (HF)³ or myocardial infarction (MI)⁴. Several modalities are available for this purpose, each with their own strengths and weaknesses. Ultrasound (US) imaging is the most widely used modality due to its great flexibility, very high temporal resolution, and relatively low cost. The adoption of US cardiac imaging in small animals has increased considerably since the advent of systems using probes with ultra-high frequency^5,6, featuring spatial resolutions below 50 µm.

Among the main disadvantages of US for fully 3D cardiac imaging is the need for linear scans along the heart axis by mounting the probe on a motorized translation stage to create a full stack of dynamic B-mode images of the whole heart⁷. Eventually, this procedure gives rise (after accurate spatial and temporal registration of the images acquired in each probe position) to a 4D image with different spatial resolutions between the in-plane and out-of-plane directions. The same problem of non-uniform spatial resolution occurs in cardiac MR (CMR),⁸ which still represents the gold standard in the functional imaging of the heart. Real isotropic 3D imaging can be instead obtained using both computed tomography (CT) and positron emission tomography (PET)⁹. PET provides a very sensitive tool in terms of image signal per quantity of injected probe (in the nanomolar range), even though it suffers from a reduced spatial resolution compared to CT, MR, or US. The main advantage of PET is its ability to display the cellular and molecular mechanisms underlying the organ's pathophysiology. For instance, a PET scan following the injection of [¹⁸F]FDG allows the reconstruction of a 3D map of the glucose metabolism in the body. By combining this with dynamic (i.e., time-resolved) data acquisition, tracer kinetic modeling can be used to calculate parametric maps of the metabolic rates of glucose uptake (MRGlu), which will provide important information about myocardial viability¹⁰.

CT requires significant volumes of external contrast agents (CA) at high concentrations (up to 400 mg of iodine per mL) to provide a measurable enhancement of the relevant tissue components (e.g., blood vs. muscle), but it excels in spatial and temporal resolution, especially when using state-of-the-art micro-CT scanners designed for small animal imaging.¹¹ A typical disease model in which the cardiac PET/CT can be applied is the experimental evaluation of myocardial infarction and heart failure and related response to therapy. A common way of inducing MI in small animals is by surgical ligation of the left anterior descending (LAD) coronary artery^12,13 and then longitudinally evaluating the progression of the disease and the cardiac remodeling in the subsequent days⁴. Nevertheless, the quantitative morphofunctional evaluation of the heart in small animals is largely applicable also for other disease models, such as the evaluation of the effect of aging on cardiac function¹⁴ or altered receptor expression in models of obesity¹⁵. The presented imaging protocol is not restricted to any given disease model and, hence, could be of the widest interest in several contexts of preclinical research with small rodents.

In this paper, we present a start-to-end experimental protocol for cardiac imaging using small animal integrated PET/CT. Even though the presented protocol is designed for a specific bimodal integrated scanner, the PET and CT parts of the described procedure could be performed independently on separate scanners from different manufacturers. In the PET/CT scanner in use, the sequence of operations is organized in a preprogrammed workflow. The main branches of each workflow are one or more acquisition protocols; each acquisition protocol can have one or more branches for specific preprocessing protocols, and in turn, each preprocessing protocol can have one or more branches for specific reconstruction protocols. Both the preparation of the animal on the imaging bed and the preparation of the external agents to be injected during the imaging procedures are described. After the completion of the image acquisition procedure, example procedures for quantitative image analysis based on commonly available software tools are provided. The main protocol is specifically designed for mouse models; even though the mouse remains the most used species in this field, we also show an adaptation of the protocol for rat imaging at the end of the main protocol. Representative results are shown for both mice and rats, demonstrating the type of output that might be expected with the described procedures. A thorough discussion is made at the end of this paper to emphasize the pros and cons of the technique, critical points, as well as how different PET radiotracers could be used with almost no modification to the preparatory and acquisition/reconstruction steps.

Protocol

Animal experiments were performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the International Guidelines on Handling of Laboratory Animals, demanded by the European Directive (Directive 86/609/EEC of 1986 and Directive 2010/63/UE) and Italian laws (D.Lgs. 26/2014).

1. Setup of the PET/CT imaging protocols and workflow

NOTE: The protocol presented here is specifically designed for cardiac imaging of mouse models. Working with rats might imply some modifications to the actual protocol, mainly because of the bigger size of the animal (roughly 10x heavier). The modifications for rat imaging are specifically mentioned in the steps; if no modifications are mentioned, then the same steps for mouse imaging can be used for rats.

1. Open the graphical user interface (GUI) of the PET/CT scanner (see Table of Materials) and create a set of new protocols (inclusive of parameters for data acquisition, preprocessing, and image reconstruction): (i) a dynamic PET scan, (ii) a low-dose CT scan for attenuation correction (CTAC) without contrast agent, and (iii) a contrast-enhanced Cine-CT scan.
   NOTE: The creation of new protocols (i.e., specific software instructions for the tomograph) for the acquisition, preprocessing, and reconstruction phases is a straightforward process; in case of trouble, the user can find more detailed information in the GUI user manual.
   1. For the PET scan, open the Protocol tab of the scanner (GUI) and create three new protocols (for acquisition, preprocessing, and reconstruction) with the following parameters:
      1. For the acquisition protocol: set 3,600 s total scan time and single bed position. Save this protocol with a proper name for subsequent import into the workflow. Do the same also for all the next protocols in the following points.
      2. For the preprocessing protocol for mouse: select a 250-750 keV energy window (EW) and enable the following corrections: radioactive decay, random coincidences, and dead time. Set the framing protocol (i.e., dynamic splitting of the raw data) as follows: 8 x 5 s, 8 x 10 s, 3 x 40 s, 2 x 60 s, 2 x 120 s, 10 x 300 s (= 3,600 s). For rat, select a 350-750 keV energy window (EW), using the same framing as for the mouse protocol.
      3. For the reconstruction protocol: select the high-quality, Monte Carlo-based 3D Ordered Subset Expectation Maximization (3D-OSEM-MC) algorithm, with 8 subsets and 8 iterations, with normalization, quantitative correction, and CT attenuation correction enabled.
   2. For the low-dose CT scan for attenuation correction (CTAC), use the following parameters:
      1. For the acquisition protocol: single frame, single bed position, full-scan; tube settings: 80 kV, low-current (low-dose); 576 views over 360°, with 34 ms of exposure time per view (20 s scan time); rotation type: continuous, sensitivity mode: high sensitivity.
      2. For the preprocessing protocol: 240 µm voxel size, transversal FOV: Rat, axial FOV: 100%.
      3. For the reconstruction protocol: filter window: smooth, voxel size: standard, enable beam hardening and ring precorrection, disable ring artifact post correction.
   3. For the contrast-enhanced gated CT scan, create three new protocols (for acquisition, preprocessing, and reconstruction) with the following settings:
      1. For the acquisition protocol for mouse: set single frame, single bed position, full-scan; tube settings: 65 kV, full-current (low-noise); 8,000 views over 360°, with 15 ms of exposure time per view (120 s scan time); rotation type: continuous, sensitivity mode: high sensitivity. For rat, set the acquisition protocol parameters as follows: 80 kV tube voltage, 16,000 views over 360°, with 12 ms exposure time per view (192 s scan time).
      2. For the preprocessing protocol for mouse: select 120 µm voxel size; transversal Field of View (FOV): Mouse; axial FOV: 50%. For rat, select a 240 µm voxel size; transversal Field of View (FOV): Rat; axial FOV: 50%.
      3. For the reconstruction protocol: filter windows: smooth, voxel size: standard; enable beam hardening and ring precorrection, disable ring artifact post correction.
   4. Open the Workflow tab in the GUI, and create a new workflow, adding the protocols just created: steps 1.1.1.1-1.1.1.3 for PET, steps 1.1.2.1. -1.1.2.3. for CTAC, and steps 1.1.3.1. -1.1.3.3. for gated CT, in the given order. In both cases, ensure that the protocols are nested with the following order: Acquisition | Preprocessing | Reconstruction.
      NOTE: Dynamic PET frames with durations <5 s to better capture the peak of the arterial input function at the beginning of the PET scan are possible but are not recommended as this can lead to noisy images with reduced quantitative accuracy. In step 1.1.2.2, we have used the "Rat" size for the transversal FOV. This is commonly used for both rats and mice in CTAC.

2. Animal preparation for PET/CT imaging

NOTE: For the present protocol, all animals were fasted overnight.

1. Anesthetize the mouse with 3%-4% (v/v) isoflurane in an induction chamber initially, and then maintain with 1%-2% (v/v) isoflurane.
2. Weigh the mouse and measure the basal glycemia to monitor the condition of the animal. To take the required blood sample, use sharp scissors and make a small cut at the tail tip, then gently massage the tail to collect a droplet of blood (~1 µL) directly on the test strip.
3. Proceed with the insertion of a venous access at the level of the caudal vein using a 29 G butterfly for mouse and 24 G for rat.
   1. To perform the cannulation technique, use concurrent heating (typically, under a heating lamp) and disinfection of the point where the needle is inserted for vasodilation of the vein. After cannulation, fix the butterfly with a silk ribbon to the tail to keep it in place during the procedure.
      NOTE: Fasting is required for [¹⁸F]FDG studies. Different tracers might involve different animal preparation, but a thorough discussion on this topic is outside of the scope of the present protocol. As far as [¹⁸F]FDG is concerned, avoiding fasting leads to very different tracer biodistribution¹⁶.
4. Turn on the anesthesia system (isoflurane 1%-2%, 0.8 L/min O₂ for mouse and 1-1.2 L/min for rat) connected to the PET-CT scanner and transfer the mouse to the bed.
5. Place the mouse in the supine position, head first, on the scanner bed of the PET-CT tomograph, putting its nose in the nose mask for anesthesia and gently blocking the head of the mouse to the mask with adhesive tape.
6. Fix the upper and lower limbs of the mouse on the scanner bed to prevent any involuntary movements during the imaging procedures, which may lead to motion artifacts.
7. Monitor the body temperature and respiration rate using a rectal probe and a respiration pillow, respectively.

3. PET tracer dose preparation

1. For mice, draw 10 MBq of [¹⁸F]FDG in a volume of 100-150 µL with an insulin syringe (1 mL). For rats, draw a higher dose of 15 MBq in 0.20-0.25 mL.
   NOTE: Avoid higher activity as the PET scanner discussed in this protocol has very high sensitivity and just requires a modest amount of activity to obtain high-quality images.
2. If the original concentration of the tracer in the vial is too high, use physiological solution (0.9% w/v NaCl) to dilute the tracer dose to a concentration of 50-100 MBq/mL.
3. Use the PET dose calibrator to measure the actual activity in the syringe. Annotate the preinjection activity and time of measurement as these values will be used later using specific input modules of the PET scanner GUI.

4. CT contrast agent preparation

1. Draw 0.2 mL per 20 g of mouse weight of iodinated lipid emulsion contrast agent into a 1 mL syringe. Limit the volume of injection to 0.5 mL of CA for heavier mice. If using iomeprol, set the injection rate for mice to 10 mL/h (~0.17 mL/min) and limit the injection volume to 0.5 mL.
   1. For rats, draw 2.3-3 mL of iomeprol, diluted to a concentration of 200 mg/mL, into a 5 mL syringe.
      NOTE: If small-animal lipid emulsion CA is not available, iomeprol with continuous injection by means of a syringe pump can be used, as discussed below.
   2. Connect the syringe to the syringe pump, and set the pump for the actual syringe size and diameter.
   3. Connect the syringe to the CA tubing and needle, and prefill the tubing with the CA.
   4. Set the injection rate to 24 mL/h (= 0.4 mL/min), limiting the injection to a maximum volume of 2 mL.
      NOTE: Using blood-pool CA based on iodinated lipid emulsion is also possible in rats, despite the relatively high cost of this procedure due to the larger volume of a single injection. If this option is preferred (e.g., to simplify the protocol by avoiding the syringe pump), the following procedure can be employed:
2. Draw 7.5 mL per kg of body weight of iodinated lipid emulsion contrast agent into a 5 mL syringe. Limit the volume of injection to 2 mL of CA also for heavier rats.

5. Animal alignment and preliminary operations before imaging

1. Upon immobilization of the animal on the imaging bed, create a new study on the tomograph GUI. Add a study name identifier in the Study name module and select the imaging workflow previously saved from the dropdown menu.
2. Select the proper anatomical part with Animal/Specimen information | Anatomical part | Cardiac and animal positioning by Animal/Specimen information | Positioning | Supine/Head first. Annotate the animal weight in grams for the corresponding module: Animal/Specimen information | Animal weight.
   NOTE: All other information in this section is optional, but it is useful to provide as much as possible of the requested information to find it in the DICOM header of the reconstruction images, thus facilitating subsequent data querying.
3. Select the radionuclide in PET Scan information | F18 for [¹⁸F]FDG studies and other ¹⁸F-labeled compounds; modify if other tracers (e.g., [¹³N]NH₃) are used. Write also the name of the tracer in the PET Scan information | Tracer name module as this name will be reported in the DICOM header upon completion of the image reconstruction.
   NOTE: The information about tracer injection time, activity, and volume is mandatory but could be provided later during the PET acquisition.
4. In the CT scan information, write all the available information regarding the contrast agent.
   NOTE: All this information is optional but could facilitate subsequent data querying if provided.
5. Press Perform scan and wait for another tab of the GUI to be opened, allowing animal positioning and the specification of other scanning options.
6. Select CT calibration type in CT calibration | Use default CT calibration.
7. In the Study preparation section, select each scanning protocol from the dropdown menu, and check the Wait for user confirmation before this scan checkbox.
   NOTE: This step is very important, as it will put the scanner in standby waiting for user input before starting the corresponding acquisition phase. For the PET scan, this will allow the synchronization of the tracer injection and actual PET scan start; for the CTAC scan, it will allow the user to close the lid (shielding) before the emission of X-rays during the CT scan (the study will be automatically aborted if the lid is open before the CT scan start); for the Cine-CT scan, this pause will allow the user to initiate the CA infusion protocol and CT data scanning with the required delay.
8. For animal positioning, switch on the Motor control module using the switch on the left pane of the GUI.
   NOTE: This will switch on the centering lasers on the animal bed and enable the manual bed alignment buttons placed on the side of the scanner.
9. Use the manual bed alignment buttons to move the chest of the animal onto the laser marks. Check carefully both the longitudinal and vertical alignment of the animal.
10. Once the animal is placed in the correct position according to the centering laser, press Switch off the laser to save the current laser-marked position, which is to be moved to the center of the PET and CT scanners during the corresponding acquisition phases. Afterward, switch off the motor control module.

6. PET scan

1. Press Start acquisition to move the animal onto the PET scanner FOV. The tail and the cannula will remain outside the FOV to allow tracer injection. The scanner will stay idle until the user presses the Continue button.
2. Prepare the syringe with the calibrated PET tracer dose.
3. Start the acquisition by pressing the Continue button and start injecting the tracer in the cannula within 5 s from the scan start (Figure 1).
   NOTE: The injection duration will be ~20-25 s.
4. Put the syringe in the PET dose calibrator to measure the residual activity in the syringe. Annotate the actual activity and the time of measurement.
5. In the Hardware monitor tab of the scanner GUI, use the Update PET tracer info button to insert the actual injected time, activity, and volume.
6. During the scan, check the animal's physiological parameters periodically.
7. During the scan, measure the glycemia as explained in step 2.2 at the following time points: 5 min, 20 min, 40 min, and 60 min after the beginning of the PET scan.
8. After the measurement of glycemia, put the test strip in the gamma counter and perform the activity measurement for 60 s. Record the actual time at which the activity measurement has been carried out, and correct for radioactive decay, taking the tracer injection time as a reference time. Convert the recorded activity values into activity concentration (Bq/mL) by considering an average blood volume of 1 µL in the glucose test strip (i.e., by using equation [1]):
   C_blood(t) = A_blood(t)/0.001 mL [Bq/mL]    (1)
   where A_blood(t) is the decay corrected measured activity of the blood sample in the test strip, expressed in Bq.
   NOTE: The PET scan start and tracer injection can be performed by the same operator by using the tomograph's mobile controlling device placed on the lateral table of the scanner close to the site of the operator during injection. Longer delays between the scan start and the beginning of the injection are allowed, but some reconstructed frame at the beginning of the dynamic sequence will remain blank. It is recommended to avoid delays greater than 10 s (i.e., leading to two blank frames with the current protocol).

Figure 1: Injection of the PET tracer. This operation is performed right after the PET scan start. The animal is inside the PET field of view (head first, with its tail visible on the operator's side). Abbreviation: PET = positron emission tomography. Please click here to view a larger version of this figure.

7. CT scans

1. Before injecting the CT contrast agent, start the CTAC scan right after closing the scanner's lid and pressing the Continue button on the GUI. At the end of this very short acquisition, employ the following procedures to ensure proper enhancement of the blood pool by injecting the CA prior to acquisition using the same vascular access employed for the injection of the PET tracer.
   1. Iodinated lipid emulsion CA:
      1. After completion of the CTAC scan, inject the iodinated lipid emulsion CA using the cannula already connected to the mouse tail vein. The typical injection duration is in the order of 30-60 s.
      2. Start the imaging right after completion of the injection. Press Continue on the scanner GUI to start the Cine-CT acquisition.
   2. Iomeprol/syringe pump:
      1. If a normal X-ray CA is used, such as iomeprol, use a syringe pump allowing slow injection at a constant rate.
      2. For mice, set the injection rate of the CA to 10 mL/h (~0.17 mL/min) by limiting the volume of injection to 0.5 mL. With this setting, stop the injection after ~3 min. For rats, set the pump to a rate of 24 mL/h (= 0.4 mL/min) and limit the volume of injection to 2 mL. With this setting, stop the injection after 5 min.
      3. Connect the needle attached to the CA tubing to the tail vein's cannula, ensuring that both the tubing and the needle are prefilled with CA.
      4. Start the injection. Close the scanner's lid and prepare for the Cine-CT scan.
      5. Press the Continue button on the tomograph's GUI after 60 s from the beginning of the injection for mice and after 90 s from the beginning of the injection for rats, so that the Cine-CT acquisition is started. The injection of CA will stop roughly at the same time as the Cine-CT scan completion for mice and upon completion for rats.
2. Upon completion of the Cine-CT scan, disconnect the animal from the physiological monitoring system and remove the tail vein cannula. Depending on the actual protocol, animals are either recovered or euthanized after the described imaging procedure. In the first case, animals are awakened in their cages in a warm environment under an infrared lamp. They are monitored until complete awakening, taking 15/30 min after gaseous anesthesia. In case of protocols requiring e.g., tissue harvesting at the end of the imaging procedure, animals are euthanized using a anesthetic overdose in an induction chamber (5% isoflurane), as per Annex VI of D.Lgs. 26/2014.
   NOTE: In the case of ¹⁸F-based radionuclides as discussed in this protocol, 24 h after tracer injection are sufficient to reach a level of residual radioactivity on the animal's body that is safe for all practical purposes.

8. Reconstruction of the cardiac 4DCT images using intrinsic cardiorespiratory gating

NOTE: Upon completion of the imaging study, the standard PET and CT reconstruction is automatically performed. Nevertheless, the reconstruction of the 4D (Cine) cardio CT sequence must be performed manually and requires some user interaction. This special type of reconstruction, mandatory for the subsequent morpho-functional cardiac CT analysis, is discussed in this section.

1. Open the cardiac gating module of the tomogragh's GUI and select the imaging study to be analyzed.
2. Select a region of interest (ROI) on the displayed animal's radiographs (Figure 2) for building a time-dependent cardiac motion curve, representing the gating signal-the kymogram. Vertically move the predrawn rectangular ROI in such a way that both the heart apex and the diaphragm are selected. Then, select Gating signal analysis. The user interface will now show the gating signal both on the time domain and frequency domain.
3. In the first frequency domain graph, select the respiratory frequency band by highlighting the first group of peaks of the frequency spectrum (refer to Figure 3 for an example spectrum).
4. In the second frequency-domain graph, select the cardiac motion frequency band, highlighting the second sharpest peak.
5. In the next phase, observe the time-domain gating signal with color markers (dots) superimposed, showing the identified respiration peaks and cardiac contraction peaks. If the marker positions fit well with the respiratory and cardiac peaks of the original gating signal, proceed to the next phase. Otherwise:
   1. If the shape of the gating signal is too different from that displayed in Figure 3, go back to step 8.2 and select another ROI.
   2. If the shape of the gating signal is reasonably similar to the one shown in Figure 3, go back to step 8.3 and step 8.4 and select different frequency bands on the gating signal spectrum.
6. In the next phase, select a minimum of four cardiac gates.
   NOTE: Typical Cine-CT reconstruction consists of 8-12 cardiac gates.
7. Select the proper respiratory window by using the dropdown menu: Respiratory window | 20%-80%.
   NOTE: This will retain 60% of the acquired data in the reconstruction, excluding the phase of peak inspiration and, thus, improving the sharpness of the reconstructed myocardial walls in each cardiac phase.
8. Perform reconstruction to convert the retrospectively gated Cine-CT images into DICOM format, ready to be imported into the software for subsequent functional analysis.

Figure 2: ROI selection tool for intrinsic gating. This image is shown in the tomograph's GUI during the Cine-CT reconstruction phase. The user must select the position of the ROI (yellow rectangle) on which the intrinsic gating signal (kymogram) is obtained from the raw CT projections. The circular-shaped object superimposed to the animal chest is the respiratory pillow used only for physiological monitoring during the study. Abbreviations: ROI = region of interest; CT = computed tomography; GUI = graphical user interface. Please click here to view a larger version of this figure.

Figure 3: Example gating signal (top frame) and corresponding frequency spectrum (center and bottom). Images obtained with the cardiac gating module of the Atrium software. The user must select the proper frequency bands for both respiratory (center frame) and cardiac motion (bottom frame). This will allow the identification of the respiratory and cardiac markers on the gating signal, which must be checked by the user before proceeding with the 4D reconstruction. Bad identification of the peaks or wrong assignment (e.g., respiratory to cardiac, or vice versa) will lead to incorrect reconstruction. The data shown were obtained from the analysis of a 4D Cine-CT scan of a healthy, adult male Wistar rat (507 g) injected with 2 mL of iomeprol, 200 mg/mL, at the rate of 0.4 mL/min for 5 min (the graph on top is zoomed in on the first 22 s of acquisition to allow better visualization of the identified cardiac and respiratory motion). Abbreviation: CT = computed tomography. Please click here to view a larger version of this figure.

9. PET cardiac analysis

NOTE: This section shows how to perform a kinetic analysis of dynamic [¹⁸F]FDG data of the small-animal left ventricle. The analysis is based on the Carimas software. The instructions below are not meant to be a replacement for the software user manual¹⁷. The procedure presented below is based on the Patlak graphical analysis of dynamic PET data¹⁸. Refer to the Discussion section for details regarding this analysis.

1. Open the DICOM images of the dynamic PET scan.
2. Select the HeartPlugin module.
3. Zoom into the image on the mouse/rat heart, and select the last time frame (or equivalently, the sum of the last three to five time frames) for which most of the blood-pool activity has been already washed out.
4. Follow the on-screen instruction to reorient the image along the principal axis of the animal heart (short axis, vertical, and horizontal long axis). Do this interactively by moving the displayed markers for the heart base and apex (Figure 4).
5. Select the Segmentation tool.
   NOTE: By default, automatic segmentation is enabled, which yields reliable results in most cases.
6. If the result of the automatic segmentation is not acceptable, refine the shape of the segmented myocardium and/or LV cavity by enabling the Manual Mode (ROI Search Disabled).
7. In the Modeling Tool, select the appropriate kinetic model to be used for dynamic PET analysis. In this case, select Graphical | Patlak to enable the Patlak plot analysis for the computation of the metabolic rate of glucose uptake (MR_Glu) for each cardiac sector.
8. In the polarmap tool, select the correct number of displayed heart segments. In this case, select 17 segments.
9. Now, press the Fit button to perform the fitting procedure of the Patlak analysis.
10. At the end of the fitting procedure, observe the displayed polar map of the K_i values (i.e., the slope of the linear regression expressed in mL/[mL × min]).
11. Using the K_i values for each sector shown in a table, calculate the MR_Glu using equation (2):
    MR_Glu = (K_i × P_Glu)/LC    (2)
    where P_Glu is a blood-sample-derived value of the plasma glucose concentration (mmol/L), and the lumped constant (LC) is an empirical coefficient used to compensate for the difference in uptake between normal glucose and FDG. See, for instance, Ng et al.²² for typical values of the lumped constant in various experimental conditions.
    NOTE: Before starting PET analysis, it is good practice to visually inspect the dynamic sequence of PET volumes within the PET analysis software tool. This is necessary to exclude macroscopic animal motion between time frames during the study. If motion is present, proper image registration (outside of the scope of this protocol) should be carried out before analysis, if possible.

Figure 4: Reorientation tool of the PET analysis software. The projection of two simple line segments in 3D space is shown on each of the three standard planes (transaxial, coronal, and sagittal). The first segment allows the user to select the heart base and apex, while the second one allows for selecting the left and right sides of the heart. This step results in a new (interpolated) PET image (bottom row), with the heart reoriented along the standard AHA representation. Images were obtained with Carimas from a healthy adult male CD-1 mouse weighing 51 g and injected with 10 MBq of [¹⁸F]FDG. Abbreviations: PET = positron emission tomography; AHA = American Heart Association; FDG = fluorodeoxyglucose. Please click here to view a larger version of this figure.

10. Cine-CT cardiac analysis

NOTE: This section shows how to perform quantitative analysis of the Cine-CT cardiac image to collect global quantitative data of the heart function. The analysis is based on the Osirix MD software. The instructions below are not meant to be a replacement for the Osirix user manual²⁴.

1. Load the DICOM images of the Cine-CT scan in the software.
2. Open the dynamic dataset with the built-in 4D viewer.
3. Using the 3D Multiplanar Reformation (MPR) tool, reorient the image data along the short axis (Figure 5).
4. Export the reoriented data to DICOM, ensuring that the entire 4D data are exported, with preserved slice thickness (same as the original) and image bit depth (16 bit per voxel)
5. Open the exported 4D MPR images using the 4D viewer.
6. Select a time frame corresponding to end-diastole. Browse through all the time frames with the time slider on the main toolbar to ensure that the correct cardiac phase is selected.
7. On this time frame, pick the closed polygon annotation tool and manually delineate the endocardial wall of the LV.
8. Do the same for 10-20 slices from the base to the apex, ensuring that all the ROIs have the same name (e.g., LVENDO).
9. On the ROI menu, select ROI Volume | Generate missing ROIs to generate the ROIs on all the short axis slices by interpolation of the manually drawn ROIs.
10. On the ROI menu, select ROI Volume | Compute Volume to calculate the volume of the ROI group with the same ROI name.
11. Browse through the time frames and select a phase corresponding to end-systole (smaller LV volume) and repeat steps 10.7-10.10 above.
12. Calculate the stroke volume (SV) and ejection fraction using equations (3) and (4):
    SV = EDV - ESV[mL]    (3)
    EF = 100 × SV/EDV [%]    (4)
    where EDV is the end-diastolic volume and ESV is the end-systolic volume.

Figure 5: Graphical interface of the multiplanar reformation tool. This tool is used for the reorientation of the Cine-CT data for subsequent functional analysis. The user shall rotate and translate the reference axes on the left side of the screen in such a way that the short-axis view of the heart is shown on the right. At the end of this procedure, the user can export the reoriented images as a DICOM file set. The images were obtained with Osirix MD and refer to a healthy adult male Wistar rat (507 g) injected with 2 mL of iomeprol, 200 mg/mL, at a rate of 0.4 mL/min for 5 min, reconstructed with Filtered BackProjection with a voxel size of 0.24 mm³. Please click here to view a larger version of this figure.

Representative Results

In this section, typical results are shown for both PET and CT analysis following the procedures described so far. Figure 6 shows the results of the automatic myocardial and LV cavity segmentation of the [¹⁸F]FDG PET scan of a control (healthy) CD-1 mouse. Even though the right ventricle is not always visible in the reconstructed images, the orientation axes based on the DICOM header can be used to correctly discriminate the interventricular septum from the other LV walls, as required for reliable identification of the standard sectors following the American Heart Association (AHA) recommendations²⁵. In the case of myocardial ischemia, a regional lowering of tracer uptake appears as a typical sign of loss of myocardial vitality. This is not necessarily correlated with reduced perfusion, which would need a different tracer (e.g., [¹³N]NH₃ or [¹⁵O]H₂O) to be visualized in PET images. Even in healthy subjects, lower reconstructed values around the apex are commonly observed in PET (see Figure 6). This might arise from a more pronounced partial volume artifact due to a (generally) thinner myocardial thickness at the apex as compared to, for example, the left wall or the septum.

Figure 6: Results of the automatic segmentation of the PET analysis software. The images were obtained with the Heart plugin of the Carimas software. Segmentation was done after standard reorientation according to AHA guidelines. The images shown refer to a healthy adult male CD-1 mouse (same as Figure 4) weighing 51 g and injected with 10 MBq of [¹⁸F]FDG, without cardiac gating, and sum the last 15 min of a 60 min PET scan. The images were reconstructed with an iterative 3D-OSEM algorithm with a voxel size of 0.85 mm³. Abbreviations: PET = positron emission tomography; AHA = American Heart Association; FDG = fluorodeoxyglucose. Please click here to view a larger version of this figure.

In Figure 7, an example of the regional K_i obtained through Patlak graphical analysis¹⁸ is shown (top left). In the frame at the bottom, the Patlak scatter plot and the corresponding results of the linear regression analysis are shown. Each point in the scatter plot represents the ratio between tissue activity concentration and plasma activity concentration at a given time t (after correction for radioactive decay), C_T(t)/C_P(t), plotted against the time integral of the plasma activity concentration from the injection time t₀ = 0 to time t. The table on the top-right frame of Figure 7 shows the values of the slope (K_i) and intercept (I_c) of the linear fit performed on each segment, along with the corresponding coefficient of determination (R²).

As far as cardiac PET is concerned, signs of bad execution of the protocol could include, but are not limited to, the following: (i) low or absent tracer uptake from the myocardium, which is typically a sign that a problem occurred during the tracer injection, such as an extravasated injection; (ii) similar problems as in the previous point if the animal temperature is too low during the PET scan (e.g., below 35 °C) and, thus, altered tracer uptake occurs; (iii) evident image blurring, which could be due to a level of anesthesia that is too low or involuntary motion.

Figure 7: Results of the Patlak graphical analysis. The images were obtained with the Heart plugin of the Carimas software. Top-left: parametric polar map of regional K_i of the LV resulting from the Patlak analysis. Top-right: average values of K_i and I_C on each myocardial segment, along with the coefficients of determination of each linear fit (R²). Bottom: scatter plot of y(t) versus x(t) (see text for details) for the selected myocardial segment (segment 1 in this example). This result refers to the myocardial PET images shown in Figure 4 and Figure 6 (healthy adult male CD-1 mouse weighing 51 g and injected with 10 MBq of [¹⁸F]FDG). Abbreviations: PET = positron emission tomography; FDG = fluorodeoxyglucose. Please click here to view a larger version of this figure.

Figure 8: Example of manual segmentation of the LV of a rat. The image refers to the same animal as shown in Figure 5 and was obtained with Osirix MD. The resulting volumetric analysis of the LV at end-diastole and end-systole is shown at the bottom. From these results, EF and SV are calculated according to equations 3 and 4. Abbreviations: EF = ejection fraction; SV = stroke volume; ROIs = regions of interest; LV = left ventricle. Please click here to view a larger version of this figure.

Figure 9: Volume rendering of the Cine-CT images. The images refer to the same rat shown in Figure 5 and Figure 8 (healthy adult male Wistar rat weighing 507 g and injected with 2 mL of iomeprol, 200 mg/mL, at the rate of 24 mL/h for 5 min, reconstructed with FBP with a voxel size of 0.24 mm³). Abbreviations: RA = right atrium; LA = left atrium; LV = left ventricle; RV = right ventricle; CT = computed tomography; FBP = Filtered BackProjection. Please click here to view a larger version of this figure.

Figure 8 and Figure 9 deal with the representative results of the Cine-CT cardiac analysis for a healthy rat. In particular, in Figure 8, the different shape and size of the LV are shown for the end-diastolic and end-systolic phases, along with the 3D reconstruction of the segmented LV volume in both phases. In this example, the calculation of the volumes according to equations 3 and 4 resulted in EDV = 0.361 mL and ESV = 0.038 mL, corresponding to a stroke volume of SV = 0.323 mL and an ejection fraction EF = 89.4%. This is in agreement with the results reported on similar protocols in the literature, showing normal EF of rats in the range of 70%-90%²⁶. Infarcted hearts might lead to a reduced EF, in the range of 50%-70% or less, depending on the lesion severity and extension of the akinetic myocardium.

The following signs of bad execution of the experiment could occur for Cine-CT images: (i) reduced or absent image contrast between the heart chambers/vessels and the myocardium; in this case, it is likely that a problem in the contrast agent injection occurred; (ii) blurred contours of the myocardial walls; in this case, a problem in the reconstruction has occurred, probably due to wrong identification of the cardiac and respiratory peaks from the intrinsic gating signal, which, in turn, might depend on bad selection of the frequency bands (Figure 3) and/or bad selection of the gating signal ROI (Figure 2); (iii) evident motion artifacts, which might be due to a level of anesthesia that is too low or involuntary motion.

In Figure 9, a volume rendering of the same rat heart is shown for both the end-diastole and end-systole. This type of visualization only allows for depicting the iodine-enhanced chambers and vessels, so their value is more qualitative than quantitative. Nevertheless, reduced motility of the myocardial walls, such as that encountered in infarcted rats, will produce volumetric images with less evident differences between the end-diastolic and end-systolic phases.

Discussion

The protocol presented in this paper focuses on a typical experimental procedure for translational cardiovascular research on small animal models of cardiac injury by using high-resolution PET/CT imaging. The presented results are indicative of the high quantitative and qualitative value of PET and Cine-CT images, providing both functional and structural information of the whole heart regarding its glucose metabolism, shape, and the dynamics of its contraction. Moreover, all the images obtained are 3D, time-resolved, and present isotropic pixel spacing; this is beneficial from the image processing point of view as it does not require pre-scan operator-dependent tasks to select specific slice orientations along the standard axes of the heart.

This paper includes a protocol based on the Patlak graphical analysis of dynamic PET data¹⁸. This type of analysis is useful to describe irreversible tracer uptake from the tissue, which is a good approximation in the case of [¹⁸F]FDG, where the effect of dephosphorylation or metabolites is generally negligible in the myocardium¹⁹. Within this approximation, the ratio between the decay-corrected tissue activity concentration C_T(t) and the decay-corrected plasma activity concentration C_P(t) can be approximated by the following equation (5):

(5)

which holds for times t, after a certain start time, t*, that must be determined empirically. In the above equation, the constant K_i represents the net influx rate from blood to tissue, while I_C is a constant involving the blood-volume fraction and the distribution volume of the tracer in the reversible compartment (i.e., the plasma). A more detailed mathematical derivation of this formula can be found elsewhere²⁰. If the time activity curves (TACs) of both the plasma and tissue are available (e.g., from a dynamic PET scan and/or plasma sampling), a 2D scatter plot can be created by plotting and for each frame time, t, so that K_i and I_C can be easily determined as the slope and intercept of the scatter plot by simple linear regression, restricted to those time points t > t* after which linearity is observed. It must be underlined that the prolonged anesthesia can impact on the metabolic rate of the myocardium²¹. For this reason, it is very important to standardize the protocol so that inter-subject variations of all the relevant physiological parameters are kept to a minimum. The procedure described in the protocol, implemented in Carimas, allows regional Patlak graphical analysis of the myocardium; we have used the activity concentration of the whole blood in the LV cavity as an approximation of the plasma input function C_P(t).

Some PET scanners might have lower spatial resolution and/or sensitivity, leading to the use of larger ROIs and consistent partial volume/spillover errors in the time activity curves (TACs) of the measurements, especially the plasmatic one used as the input function (IF). In this case, the analysis protocol can be modified by creating a hybrid IF based on the image values in the early phase after injection and on blood sample activity concentration (see protocol step 6.8) in the late phase (>20 min). The corrected points of the hybrid IF can be calculated by interpolation, as shown by Shoghi et al.²³. Within Carimas, it is possible to export the raw TACs of each myocardial segment, correct the arterial TAC, and load them again to perform Patlak analysis directly on the corrected curves. Due to the complexity of the operation required, we have not provided specific protocol operations to do that, as the results obtained in the case described in this protocol have a good level of reproducibility for most applications.

A possible application of the presented protocol is in small animal models of myocardial infarction. To avoid restrictions to such a specific field of imaging research, we did not add a specific protocol instruction for the induction of MI or other types of cardiovascular diseases. Detailed surgical procedures can be found elsewhere in the literature^12,13, and they have been successfully applied in our group with the goal of depicting complementary information for both regional perfusion defects and ischemia-induced angiogenesis⁴. Nevertheless, the PET/CT imaging protocol presented in this paper can be useful in a variety of study designs when cardiac metabolism, function, and/or morphology are of concern, including, but not limited to, metabolic diseases²⁷, response to therapy and/or to different diets²⁸, and radiation-induced injuries²⁹. Moreover, this type of investigation can be useful when validating new molecular probes for monitoring cardiac remodeling and neovascularization in correlation with global and regional cardiac function and morphology ⁴.

Here, we have discussed a typical PET image acquisition and analysis focused on the quantification of myocardial regional glucose uptake by means of [¹⁸F]FDG; for myocardial infarction imaging, for instance, this is useful and widely adopted for measuring the myocardial viability²⁶ as complementary information about perfusion, which instead requires different tracers. Moreover, [¹⁸F]FDG is the most widely available tracer in PET imaging, and hence, we decided to tailor this protocol to this tracer to increase its applicability. With minor modifications to the analysis workflow, the same procedure can be used to quantify, for example, the regional myocardial blood flow (MBF), employing [¹³N]NH₃ or [¹⁵O]H₂O as blood flow tracers³⁰.

In these cases, the PET acquisition protocol requires minor changes, considering the different radionuclide decay times of ¹³N (T_1/2 = 9.97 min) and ¹⁵O (T_1/2 = 2.04 min) with respect to ¹⁸F (T_1/2 = 109.8 min). Further, appropriate kinetic models must be used instead of the one presented in this paper, which is commonly available in most quantitative image processing software for PET analysis; besides these points, the experimental procedure presented in this protocol is mostly suitable for other types of experimental investigation focused on the hearts of small animals. Although the protocol was specifically designed for cardiac imaging of mouse models, working with rats might imply some modifications to the actual protocol, mainly because of the bigger size of the animal (~10x heavier). However, additional information has been added to the protocol to indicate the required modifications for rat imaging for the sake of simplicity.

An advantage of the presented protocol is that it does not require the use of ECG probes on the animal, as the PET study can be reliably conducted without gating, and the CT study uses intrinsic (sensorless) retrospective gating. The algorithm at the base of the intrinsic gating software is based on the work of Dinkel et al.³¹. This method shows very high agreement with ECG-based (extrinsic) cardiac gating and may even potentially be better in the case of arrhythmias due to the dissociation of mechanical and electrical events³¹. Even though intrinsic gating could be implemented in fully automated workflows³², this protocol is based on an interactive method implemented in the IRIS CT scanner, giving more flexibility in the choice of the parameters. As discussed, minor adaptations of the procedures are required when using rats instead of mice, primarily regarding the injected doses, the need for attenuation correction (CTAC) scans when using larger animals, as well as some differences between the types of CT contrast agents. Regarding this last point, the usage of iodine-rich oil-in-water lipid emulsions on rats is also reported in the technical notes of small animal CA vendors. Due to the relatively large injection volumes involved, the relatively higher cost, and the less widespread availability of these specialized contrast agents, we have also presented a modification of the protocol based on commonly available vascular contrast agents, such as iomeprol, which is broadly applicable in clinical settings. Due to the very fast clearance of such standard vascular agents, a motorized injection pump allowing slow continuous injection is required in this case.

Limitations of the method
The applicability of the presented PET/CT protocols relies on the availability of instrumentation that is generally less widespread and more expensive than other techniques (primarily US echocardiography), even though the contextual information about structure, function, and metabolism cannot be achieved by any other technique with the same sensitivity and flexibility on the choice of the molecular probe. However, the successful completion of the entire preparation/acquisition/analysis workflow with this methodology requires a strong collaboration between several professional figures, including biologists, veterinary doctors, chemists, physicists, and bioengineers. This is even more true when non-standard PET tracers are used, implying efforts in both radiosynthesis and mathematical modeling, as well as in the adaptation of the analysis software for correct and reliable quantification^33,34,35.

In protocol section 9, we have described a very simple quantification procedure by using an image-derived input function (IDIF), pointing out that a mixed approach using IDIF and blood sample-derived IF for late frames can give better results. It must be noted that using activity measured from whole (venous) blood taken from the tail is considered a reliable approximation in [¹⁸F]FDG, but it requires further corrections for the activity of metabolites in the case of different tracers^36,37. One of the most critical points of the entire protocol is the intravenous cannulation, giving venous access for injecting both the radioactive tracer for the PET scan and the iodinated contrast agent for the CT scan. Unsuccessfully carrying out this critical step results in useless images, as the effective quantity of circulating PET tracer or CT CA might be lower than required. Expert personnel with specific training for tail vein injection must be involved in this procedure to provide reliable results.

A drawback of CT for dynamic cardiac imaging is its relatively lower temporal resolution as compared to US and MRI, even though 3D cardiac imaging with ultrasound requires the use of a motorized translating stage for the probe and subsequent image registration to get correct results. The need for consistent volumes of CA to be injected for the correct discrimination of blood and myocardium in reconstructed images is one of the main concerns due to the intrinsic low sensitivity of the methodology. In this protocol, we have limited the volume of injection of CA for CT studies to 0.5 mL in mice and 2 mL in rats, using continuous infusion for 3 min at 10 mL/h in mice and for 5 min at 24 mL/h in rats. We have observed that these rates and volumes of injections are well tolerated by the animals. The quantities described here are in line with or are smaller than equivalent protocols found in the literature.

Nahrendorf et al. described a Cine-CT protocol for the depiction of murine myocardial infarction, involving a basal (pre-scan) bolus injection of 0.2 mL of oil-in-water lipid emulsion blood pool CA followed by a continuous injection of iomeprol at 1 mL/h for 1 h³⁸. Badea et al. compared a similar cardiac Cine-CT protocol based on a 1 h infusion of Isovue 370 (iopamidol) with a bolus injection of 0.5 mL/25 g of body weight of Fenestra VC (oil-in-water lipid emulsion), finding better results in terms of image contrast in the second case³⁹. The same manufacturer of the Fenestra VC contrast agent reported a recommended volume of injection of 0.4 mL/20 g of body weight for vascular imaging with micro-CT⁴⁰. However, novel CAs with higher density such as eXIA 160 XL, MVivo Au, Aurovist 15 nm, or Exitron nano 12000 have recently entered the preclinical market and have the potential to reduce the injection volumes in cardiac micro-CT protocols. Nebuloni et al. conducted an extensive characterization of such CAs⁴¹. The radiation dose in gated CT is another common concern for longitudinal studies; in this case, the maximum dose for the described Cine-CT protocol is below 200 mGy for both mice and rats, as estimated on the basis of previous dosimetric characterization of our CT scanner⁴². This is roughly 5x lower than the reported dose in the literature for 4D cardiac CT scans^38,39 and 30x lower than the average lethal dose for total body irradiation of small animals, estimated at 6 Gy⁴³.

Applicability of the protocol to different instrumentation and software
Even though the specific instructions presented in this protocol are unavoidably tailored for a specific PET/CT tomograph, the imaging tasks presented here can be adapted to different imaging systems. Regarding the PET section of this protocol, all state-of-art PET or PET/CT systems designed for small animal research have performance requirements (in terms of spatial and temporal resolution) suitable to carry out the protocol. As far as cardiac CT is concerned, the protocol might change depending on the specific cardio-respiratory gating system employed (e.g., extrinsic or intrinsic). The readers can refer to recent review articles and book chapters for a thorough discussion of current PET, CT, or PET/CT systems' capabilities^44,45,46. Of note, the CT and PET protocols presented in this paper can be performed independently, based on the capabilities and peculiarities of the tomographic instrumentation in use. We, therefore, believe that the presented procedures can provide a useful reference for any practitioner interested in conducting for the first time, a cardiac PET/CT study on small animals.

Any user with sufficient skills in the general protocol setup of their own PET/CT tomograph should be able to implement the required adaptations on the presented method to get equivalent results in their lab. The same arguments can be used for the section dedicated to image analysis. A thorough list of all the available software packages for cardiac PET and cardiac CT analysis is outside of the purpose of this paper. However, many other comparable software packages use a similar methodology for polar map generation and regional tracer kinetic analysis. The reader can refer to Wang et al.⁴⁷ and references within for the task of PET quantification and to pertinent research articles^48,49,50 for 4D CT quantification. In this case, we have decided to focus this protocol on Carimas^51,52,53,54 and OsiriX^55,56,57,58 for the quantitative analysis of cardiac PET and CT images, respectively. Due to the widespread use of these tools, we believe that this choice could be helpful in increasing the interest of the research community in the implementation and application of the presented methods, as compared to a discussion focused on closed, commercial, and scanner-specific analysis tools provided by some PET and CT scanner manufacturers.

Modifications to the quantitative image analysis protocol
The example results shown here are just a simple outcome of a straightforward quantitative analysis task, which might be considered sufficient for most practical purposes in translational cardiovascular research experiments focused on small animal models of cardiac injury. However, many more analysis options are possible, starting from the DICOM images resulting from the acquisition/reconstruction protocol described in this paper. For instance, one might be interested in applying different compartmental models instead of the Patlak graphical analysis from dynamic [¹⁸F]FDG-PET data^59,60,61. Moreover, the analysis of the cardiac function based on 4D Cine-CT images shown in this protocol was only global for the entire LV, but several different (mainly commercial) software allow users to perform strain analysis and regional wall motion, wall thickening, and regional EF analysis from the same images⁴⁹. Nevertheless, we believe that the examples shown here represent a good starting point for more in-depth post-processing and quantitative tasks.

Materials

Name	Company	Catalog Number	Comments
0.9% sterile saline	Fresenius Kabi		0.9% sodium chloride for injection
1025L Physiological Monitoring	Small Animal Instruments		Physiological monitoring system for small animal imaging
5 mL syringes	Artsana		Syringes with needle for injection of PET tracer
Atomlab 500	Else Nuclear		PET Dose calibrator
Atrium software	Inviscan	Version 1.5.5	PET/CT operating software
Butterfly catheters	Delta Med		27.5 G needle
Carimas software	Turku PET Center	Version 2.10	Image analysis software
Fenestra VC	Medilumine		Lipid emulsion iodinated contrast agent for small animals
Heat lamp			Heat lamp with clamp and switch
Insulin syringes	Artsana		Syringes with needle for injection of CT CA
Iomeron 400 mgI/mL	Bracco		Iomeprol, vascular contrast agent
IRIS PET/CT	Inviscan		PET/CT scanner for small animals
Isoflurane	Zoetis		Inhalation anesthetic, 250 mL
OneTouch Glucometer	Johnson&Johnson Medical		Glucose meter kit
Osirix MD software	Pixmeo	Version 11	Image analysis software
Oxygen	Air liquide		Compressed gas
Rectal probe for 1025L	Small Animal Instruments		Rectal probe with cable for SAII 1025L systems
Respiratory sensor for 1025L	Small Animal Instruments		Respiratory pillow with tubings for SAII 1025L systems
TJ-3A syringe pump	Longer		Motorized syringe pump for CT CA injection