Comparison of 4D-MicroSPECT and MicroCT for Murine Cardiac Function
- First Online:
- Cite this article as:
- Befera, N.T., Badea, C.T. & Johnson, G.A. Mol Imaging Biol (2014) 16: 235. doi:10.1007/s11307-013-0686-z
- 376 Views
The objective of this study was to compare a new generation of four-dimensional micro-single photon emission computed tomography (microSPECT) with microCT for the quantitative in vivo assessment of murine cardiac function.
Four-dimensional isotropic cardiac images were acquired from anesthetized normal C57BL/6 mice with either microSPECT (n = 6) or microCT (n = 6). One additional mouse with myocardial infarction (MI) was scanned with both modalities. Prior to imaging, mice were injected with either technetium tetrofosmin for microSPECT or a liposomal blood pool contrast agent for microCT. Segmentation of the left ventricle (LV) was performed using Vitrea (Vital Images) software, to derive global and regional function.
Measures of global LV function between microSPECT and microCT groups were comparable (e.g., ejection fraction = 71 ± 6 % microSPECT and 68 ± 4 % microCT). Regional functional indices (wall motion, wall thickening, regional ejection fraction) were also similar for the two modalities. In the mouse with MI, microSPECT identified a large perfusion defect that was not evident with microCT.
Despite lower spatial resolution, microSPECT was comparable to microCT in the quantitative evaluation of cardiac function. MicroSPECT offers an advantage over microCT in the ability to evaluate simultaneously myocardial radiotracer distribution and function, simultaneously. MicroSPECT should be considered as an alternative to microCT and magnetic resonance for preclinical cardiac imaging in the mouse.
Key wordsMurine cardiac functional imagingMolecular imagingMicroSPECTMicroCTDynamic microSPECTMicro-CTFour-dimensional (4D) cardiac imaging99mTc tetrofosminMiceCardiovascular imaging
In vivo cardiac imaging in mouse models has proven useful in the study of human cardiac disease as well as in the development of potential treatment strategies [1–5]. Given the small size of the mouse heart (long axis is approximately 7 mm) and rapid heart rate (450–600 beats per min) , murine cardiac imaging is challenging because both high spatial and temporal resolutions are required.
Despite these challenges, multiple preclinical imaging modalities are currently available to measure murine cardiac function. Successful studies have been previously reported for both magnetic resonance (MR) [7–12] and microCT [13–17]. Fewer studies in the mouse, however, have explored single photon emission computed tomography (SPECT) as a potential tool for evaluating left ventricular function [18–21]. This is likely due to the typically lower spatial resolution of older-generation microSPECT systems. Recently, subhalf-millimeter resolution has been achieved with a new generation of microSPECT scanners using multi-pinhole collimation . To date, no studies have evaluated high-resolution microSPECT for cardiac functional imaging in the mouse or compared microSPECT with other modalities for use in preclinical cardiac imaging. Consequently, the goal of this study was to evaluate microSPECT as a tool for measuring murine cardiac function by comparing it directly with microCT.
While MR has been considered the “gold standard” for cardiac imaging in the mouse, both microSPECT and microCT may offer alternatives and even some advantages. In this regard, it is important to inquire—what are the tradeoffs for assessment of cardiac function? Clearly, there are marked differences in several imaging parameters between microSPECT and microCT, including spatial resolution, voxel volume, and contrast-to-noise ratio (CNR) [3, 8–10, 14, 15, 17–19], but it is not clear how these differences impact the volumetric analysis of cardiac function. Cardiac imaging with microSPECT is attractive as a potential alternative to microCT and MR in that it permits both functional analysis and quantitative evaluation of myocardial radiotracer distribution from a single acquisition. The limited resolution of previous generations of microSPECT systems (0.6–1.0 mm) [18, 19] also made them a poor option for evaluating cardiac anatomy. However, the spatial resolution of the microSPECT system evaluated here (at 350 μm) presents the possibility of simultaneous anatomic, functional, and molecular imaging.
Here, we first characterize the imaging protocols in control C57BL/6 mice for microSPECT and microCT and, using the same functional analysis, compare measures of both global and regional left ventricular function. We then compare four-dimensional (4D) microSPECT and microCT datasets obtained from a mouse with myocardial infarction. The advantages and disadvantages of microSPECT and microCT will be discussed.
Materials and Methods
In Vivo Experiments in Mice
All animal imaging studies were approved by the institutional Subcommittee on Research Animal Care of the Duke University Institutional Animal Care and Use Committee. C57BL/6 mice were separated into microSPECT (n = 6) and microCT (n = 6) groups. One additional C57BL/6 mouse with myocardial infarction was imaged with both modalities in a 24-h period.
MicroSPECT acquisitions were performed with a commercially available U-SPECTII/CT system fitted with a 0.35-mm multi-pinhole collimator (MILabs, Utrecht, The Netherlands). This isotropic voxel volume (42 nl) is more than five times smaller than the voxels of previous-generation microSPECT systems .
During image acquisition, animals were placed prone on a custom heated animal bed with integrated electrocardiogram (ECG) and respiratory monitoring (MILabs, Utrecht, The Netherlands). Body temperature was maintained at 38 °C. Mice were maintained under general anesthesia throughout the imaging procedure with 1.5 % isoflurane, delivered via a nose cone at a rate of 0.4 l/min. Anesthetized mice were injected via a tail vein catheter with 185–370 MBq of technetium (99mTc) tetrofosmin (GE Healthcare, Arlington Heights, IL) prior to image acquisition. Care was taken to minimize injection volume by using a delivery system developed in our laboratory, which eliminates the need for a post-injection saline flush and permits total injected volumes as low as 50–100 μl for the radiotracer dose range used in this study. Briefly, a modified and blunted 24-guage needle (Hospira, Inc., Lake Forest, IL) is fitted to a short length of PE-50 polyethylene tubing (Intramedic®, Becton Dickinson and Co., Sparks, MD), which is then attached to a 1-cm3 syringe (Becton Dickinson & Co., Franklin Lakes, NJ). Radiotracer is drawn into the modified needle, which is then inserted through the plastic hub of the tail vein catheter to form a watertight seal, thereby bypassing dead space within the catheter hub. Radiotracer can then be injected, and the modified needle removed, without the need to clear indwelling saline or perform a saline flush afterward.
Immediately following radiotracer injection, microSPECT images were acquired over 120 min. The field of view was adjusted to the margins of the heart using orthogonal radiographs generated by the integrated microCT unit. Data were acquired in list mode and reconstructed with retrospective cardiac gating (10-time bins per cardiac cycle) using the built-in U-SPECTII/CT reconstruction software; this software utilizes the pixel-based ordered subset expectation maximization iterative reconstruction algorithm (6 iterations and 16 subsets, reconstructed at a voxel size of 0.125 mm) . Reconstructed images were viewed, and the Gaussian filter was applied (0.4 full width at half maximum) using PMOD v.3.3 biomedical image quantification software (PMOD Technologies Ltd., Zurich, Switzerland); images were then transferred to the separate Vitrea® (Vital Images, Inc., Minnetonka, MN) workstation for further analysis.
MicroCT data acquisitions were performed with a dual-source microCT system developed in our laboratory explicitly for dynamic applications, which has been described in detail elsewhere . This system offers the highest spatial resolution currently available for 4D cardiac microCT imaging with isotropic spatial resolution of 217 μm as measured at 10 % modulation transfer function and corresponding to 2.3 lp/mm [13–15, 25]. The system contains two G-297 X-ray tubes (Varian Medical Systems, Palo Alto, CA) with 0.3/0.8 mm focal spot size, two Epsilon High Frequency X-ray generators by EMD Technologies (Quebec, Canada), and two CCD-based detectors with a Gd2O2S phosphor (XDI-VHR 2 Photonic Science, East Sussex, UK) with 22-μm pixels, which we typically bin to 88 μm. The X-ray tubes and detectors are arranged orthogonally. The data acquisition is controlled by a sequencing application written in LabVIEW (National Instruments, Austin, TX). To limit motion blur, we used pulsed X-rays at 80 kVp and 100 mA, with a 10-ms exposure time.
During imaging with microCT, mice were anesthetized with isoflurane (1.5 %) mixed with 40 % oxygen and balanced with nitrogen, delivered via a nose cone at 0.4 l/min, and were maintained under general anesthesia throughout the entire imaging procedure. ECG was monitored with electrodes (Blue Sensor, Medicotest, UK) taped to the front footpads. Body temperature was maintained at 38 °C with heat lamps connected to a rectal probe and feedback controller (Digi-Sense®, Cole-Parmer, Chicago, IL). A pneumatic pillow on the thorax was used to monitor respiration.
To provide the necessary blood/tissue contrast, we used a liposomal blood pool contrast agent containing 123 mg iodine per ml . The liposomes were delivered by injection via a tail vein catheter, at a dose of 0.012 ml/g body weight. To avoid volume overload, care was taken to ensure a total injection volume of <0.3 ml. Following administration of the liposomal agent, contrast between blood and myocardium was approximately 350–450 Hounsfield units (HU).
Immediately after injection of the liposomal contrast agent, we performed 4D microCT sampling using retrospective cardiorespiratory gating . Projection images were acquired at a rapid and constant rate of 20 projections/s without waiting for cardiac and respiratory coincidence. Respiratory signals and ECG were monitored and saved in synchrony with the acquisition of the projections. Sampling involved five rotations of the animal, with 450 projections acquired per each full rotation. A total of 2,250 projections were acquired, with acquisition time between 5 and 10 min.
The procedures for post-sampling processing and reconstruction based on retrospective gating data have been described in detail elsewhere  but are summarized here. A MATLAB (MathWorks, Natick, MA) script was used to sort the projection images in association with angle and cardiorespiratory phase information. This script detects the R peaks in the ECG signal and the maxima in the respiratory signal. Each cardiac cycle is divided into 10 intervals (each equal to 10 % of the RR interval). Each projection is temporally registered with the ECG and respiration signal and then assigned to a certain cardiac phase by calculating the temporal distance between the projection sampling time and the nearest R peak and dividing that distance by the RR interval. A similar operation is used to label projections with information related to respiratory phase. This information is stored and used during reconstruction.
Reconstruction of each set is performed with a filtered back projection Feldkamp algorithm . Streak artifacts are often seen in the reconstructed images acquired with retrospective gating. This occurs because, in retrospective gating, the angular distribution of the projections is irregular. To overcome this problem, we generated a synthetic set of projections with a regular angular distribution by using a weighted interpolation of projections over all angles and all cardiac and respiratory phases . To reduce noise and any remaining artifacts, this method also employs multi-dimensional bilateral filtration.
MicroCT imaging resulted in 10 separate 3D datasets that correspond to 10 phases of the cardiac cycle, each with a matrix size of 512 × 512 × 300, a reconstructed voxel size of 88 μm, and a voxel volume of 0.68 nl.
Imaging Myocardial Infarction
To further compare microCT and microSPECT, we imaged a mouse with myocardial infarction (MI) at 14 days after left anterior descending (LAD) coronary artery ligation procedure. An additional C57BL/6 mouse was used to create a model of myocardial infarction induced by coronary artery ligation, as described elsewhere . The mouse was maintained under general anesthesia throughout the surgical procedure, as detailed above. A thoracotomy was performed in the fourth left intercostal space. The left ventricle (LV) was visualized, and the LAD coronary artery was permanently ligated with a monofilament nylon 8-0 suture at the level of the left atrial appendage. This mouse was imaged with microSPECT and microCT according to the protocols described previously. MicroCT imaging was performed first, followed by microSPECT imaging 24 h later.
To estimate absorbed radiation dose to the animal for microSPECT, we used the most recent version of MIRD software (OLINDA/EXM) to compute the radiation dose to a 25-g tissue-equivalent sphere . Taking into account both physical decay and fecal/urinary excretion, we estimated the self-dose to a 25-g mouse to be 8.23 rad per mCi. The radiation dose in microCT was estimated using MOSFET detectors .
Comparison of regional cardiac function was accomplished using the method reported by Constantinides et al. . For this analysis, both endocardium and epicardium are semiautomatically segmented using Vitrea.
Based on this analysis, in both study groups and for the single animal with myocardial infarction, bull’s-eye plots of regional left ventricular function were generated. These plots divide the left ventricle into 17 segments according to the American Heart Association guidelines . Sector 17 was omitted from the analysis as no data could be obtained for this segment for either modality. This can be explained by the fact that in the Vitrea software, if the inner and outer apex planes which comprise sector 17 are very close together (as would be expected in the mouse), this segment will be displayed as a blank circle in the 17-segment polar map . The mean values in each of the segments were used to compute statistics over multiple mice in both microSPECT and microCT groups and to construct composite bull’s-eye plots for microSPECT and microCT groups using a polar data plot script in MATLAB . Mean global and regional functional indices from microCT and microSPECT groups were compared using a two-tailed unpaired t test and p values of ≤0.05 were considered to be statistically significant.
Comparison of microSPECT and microCT imaging parameters, advantages, and limitations
350 μm3 × 350 μm3 × 350 μm3
217 μm3 × 217 μm3 × 217 μm3
5.09 ± 1.6
8.82 ± 1.9
•Evaluates both function and perfusion in a single acquisition
•Quantifies radiotracer uptake in tissues
•Quantitative measurement of tissue density
•Detects myocardial ischemia/infarction
•Lower radiation dose vs SPECT
•Nanomolar sensitivity in detection of molecular probes 
•Higher spatial resolution and CNR compared to SPECT
•Longer acquisition time
•Requires contrast agent
•Unable to detect myocardial perfusion defectsc
In this study, microSPECT group mice received 185–370 MBq of 99mTc tetrofosmin approximately 5 min prior to scanning, which resulted in an approximate absorbed dose to the animal of 400–800 mGy, compared with approximately 360 mGy for microCT. The lethal dose (LD)50/30 for mice is reported to be approximately 7 Gy [36, 37].
In this study, we have directly compared microSPECT with microCT as an imaging tool for the evaluation of murine cardiac function. While MR has been widely used for functional cardiac imaging in the mouse with both 2D and 3D acquisitions, microCT has been employed less commonly, primarily because commercial systems utilizing micro-focal spot tubes have very low output and require significant integration time/view to achieve reasonable contrast-to-noise ratio. The system we used for comparison here has been designed explicitly for rodent cardiac imaging and represents what we believe to be the state of the art for cardiac microCT. Both microSPECT and microCT are valuable alternatives to MR for preclinical cardiac imaging and offer distinct advantages. For example, both the microSPECT and microCT systems presented in this study produce 3D isotropic datasets. In contrast, preclinical MR, while similar to microCT in spatial resolution and CNR, does not typically produce isotropic data. Indeed, the vast majority of work in imaging the mouse heart with MR has been done by imaging only a few slices through the heart, with a small number of publications that present a true 3D imaging technique [3, 8, 38]. MR application to murine cardiac functional imaging has also been limited given that acquisition times are long, and segmentation is usually not automated .
By directly comparing microSPECT with microCT, we found, despite a marked difference in spatial resolution and CNR, an excellent agreement in measured cardiac function between the two modalities (Fig. 5). It should be noted that this was the case despite a fundamental difference in the contrast-generating material used, i.e., liposomal iodinated contrast agent in CT remains largely in the vascular space, while 99mTc tetrofosmin leaves the blood pool and is taken up by the myocardium. The agreement in calculated mean cardiac output between microSPECT and microCT groups was good, differing by approximately 8 %, i.e., 0.8 ml/min (9.8 ± 2.2 and 9.0 ± 1.7 ml/min for microSPECT and microCT); global LV ejection fraction was also well matched, at 71 ± 6 % for microSPECT and 68 ± 4 % for microCT (a relative difference of approximately 5 %). Not surprisingly, the higher-resolution microCT gave a slightly tighter standard deviation for all indices (it was 4 % in microCT vs 6 % in microSPECT for left ventricular EF).
Quantitative measures of mean wall motion, wall thickening, and regional ejection fraction were also closely matched, demonstrating similar patterns of transmural variation in the two groups in most sectors (Fig. 6). This agreement is supported by the absence of statistically significant differences in all but a few cases. The few significant differences that are seen may be a consequence of volume averaging in the lower-resolution microSPECT images.
The mean EF and CO values we obtained in both study groups were comparable to what is reported elsewhere [8–10, 17–19]. Our results for EDV and ESV are similar to those obtained by Dawson et al.  in a study comparing 3D echocardiography with MR for the evaluation of LV function. Our study also showed good internal agreement between calculated EDV and ESV values in the two modality groups. There were differences, however, in our calculated EDV and ESV values compared with some prior studies [3, 8, 14, 17] which may have resulted from differences in injected contrast agent volume and anesthetic conditions. In prior experiments, we have observed that injected volumes of >300 μl result in hemodynamic consequences including reduced EF and increased EDV and ESV (unpublished). This 300 μl volume limit can easily be exceeded if care is not taken to reduce or eliminate saline flushes during catheter insertion and following contrast injection. We have developed procedures for contrast (microCT) and radiotracer (microSPECT) injection which minimize saline flush volume, ensuring that the total volume of injection is <300 μl (see “MicroSPECT Imaging” section). Differences in our calculated LV volumes compared with published MR data [3, 8, 11] could also be related to variations in the method used for volumetric analysis. MR volumetric measurements were typically calculated by manual threshold segmentation, whereas our study utilized semiautomated segmentation with the Vitrea LV functional analysis software .
One of the relative advantages of microSPECT over microCT was evident when both modalities were used to image a mouse with a surgically induced myocardial infarction. MicroSPECT readily identified a large perfusion defect consistent with MI involving the cardiac apex and anterolateral LV wall, which could not be seen by microCT (Fig. 7(a–c)). Although both microCT and microSPECT did show decreased LV wall motion consistent with the presence of an infarct (see arrowheads on wall motion bull’s-eye plots in Fig. 7(d)), only microSPECT revealed the actual margins of the infarct and associated perfusion deficits. We note, however, that other microCT techniques for imaging myocardial infarction are possible. For example, MI in the mouse has been evaluated via delayed hyper-enhancement  using conventional contrast agents such as Isovue 370 (Bracco Diagnostics, Inc., Monroe Township, NJ), although this is less common. Alternatively, eXIA 160 (Binitio Biomedical, Inc., Ottawa, Ontario, Canada), an aqueous colloidal polydisperse contrast agent, has been used recently by our group for imaging of myocardial infarction with dual-energy microCT . The 2D and 3D MR techniques have also been used less commonly to identify myocardial infarction in pig and mouse models [41, 42]. When compared with these alternative microCT and MR techniques, microSPECT still offers a relative advantage in imaging myocardial infarction in that both function and radiotracer distribution data can be obtained from a single acquisition and without the need for additional specialized contrast agents.
MicroSPECT also has several limitations when compared with microCT. Comparison of imaging parameters in the two modalities (Table 1) reveals a marked disparity in spatial resolution and voxel volume between microSPECT and microCT. Despite the use of a high-resolution 0.35-mm multi-pinhole collimator in this study, microSPECT voxel volume was still four times larger than voxel volumes in the comparison microCT datasets (Table 1). This difference, however, did not appear to be a factor in volumetric segmentation of the left ventricle using the Vitrea software. The lower CNR of the microSPECT images (5.09 ± 1.66 for SPECT vs 8.82 ± 1.89 for microCT) also did not appear to impair functional analysis of the 4D microSPECT datasets.
Time required for imaging is another important consideration when comparing modalities. Acquisition time for microSPECT (120 min) was longer than the 30–60 min scan times reported in published MR studies [3, 8, 10, 38] and was considerably longer than microCT acquisition time (5–10 min) and could therefore become a limitation in throughput. It should be noted, however, that the microSPECT imaging protocol was not optimized for speed, but rather for image quality, and that 4D microSPECT datasets which are adequate for functional analysis were possible over a range of injected radiotracer activity (185–370 MBq). It is likely that microSPECT scan time could be reduced without negatively impacting functional analysis. Further experimentation is necessary to determine the minimum combination of radiotracer dose/scan time required to produce microSPECT images adequate for volumetric segmentation in the Vitrea software.
A limitation of both microSPECT and microCT when compared with other imaging modalities, such as MR, is the use of ionizing radiation. The absorbed dose associated with the methods in this study is not negligible but represents between 9 and 18 times less than the mouse LD50/30 in the case of microSPECT and approximately 19 times less than the mouse LD50/30 in the case of microCT . Note that while this level of radiation exposure would have no effect on cardiac imaging, it could potentially become an issue in cancer studies by affecting the progression of some tumors. For microSPECT, the dose of radiopharmaceutical selected can vary widely depending on the size of the subject and the particular signal-to-noise ratio requirements of the experiment. In this study, the dose was deliberately chosen at the high end of the spectrum to provide the highest quality of images and to make comparisons between microCT and microSPECT more appropriate.
Despite its limitations, microSPECT has several advantages over microCT in the evaluation of murine cardiac function (Table 1). MicroSPECT enables quantitative analysis of both cardiac function and myocardial radiotracer distribution from a single 4D acquisition, without the need for multiple contrast agents or additional acquisition protocols. MicroSPECT readily identifies areas of ischemia or infarction which are not apparent with standard microCT imaging alone. These benefits should have practical applications in a number of preclinical experimental protocols involving mouse models of cardiac disease.
We have shown that in the mouse, microSPECT is similar to microCT in its ability to evaluate cardiac function. We compared our microSPECT system with the highest-resolution 4D microCT currently available and found no significant differences between global LV functional indices in two separate but similar groups of mice. Regional mean functional indices (wall motion, wall thickening, rEF) were also similar with no significant differences in these parameters in the majority of LV sectors.
When compared with microCT, microSPECT offers several advantages with respect to murine cardiac imaging. Specifically, microSPECT provides the capability, with a single acquisition, to quantify radiotracer uptake in the myocardium and to identify global or regional ischemia. MicroSPECT also detects molecular probes with a 103 greater sensitivity than with currently available MR techniques [3, 5]. Given that microSPECT provides these added benefits, while also delivering a functional analysis equivalent to state-of-the-art microCT, microSPECT should be considered as a valid alternative to other preclinical imaging modalities when studying mouse models of cardiac disease.
All work was performed by the Duke Center for In Vivo Microscopy, an NIH/NIBIB Biomedical Technology Resource Center (P41 EB015897). Special thanks go to Yi Qi for helping with animal setup and to Sidney Simon and Sally Zimney for their editorial assistance. Liposomal contrast agent for microCT has been provided by Dr. Ketan Ghaghada (Texas Children’s Hospital).
Conflict of Interest
The authors declare that they have no conflicts of interest.