Cardiac MRI is regarded as the criterion-referenced standard for examination of the heart with respect to accuracy and reproducibility of volumes, mass, and wall motion in human medicine.1 Moreover, cardiac MRI provides excellent spatial and temporal resolution2 and a detailed view of soft tissue.3 In veterinary medicine, cardiac MRI has proven to be a reliable imaging modality for the visual examination of morphological and anatomic heart structures and the associated great vessels in both physiologic and pathological conditions.4,5 Furthermore, cardiac MRI can be used to examine myocardial function of the heart.
Because it provides high resolution and good detail of soft tissues, cardiac MRI has been adopted as a reference technique for the assessment of LV function variables in many human studies.2,6–8 The use of cardiac MRI to evaluate LV function variables in dogs has been investigated.9–11 Nevertheless, in veterinary practice, its use in dogs3 is still limited because it is a time-consuming and expensive modality as a result of the costs of the cardiac MRI device and operating software. Additionally, an experienced operator is needed to obtain images and analyze acquired images.
The imaging modality cardiac CT also yields a good depiction of the cardiac structures.6,12 In a recent study,13 investigators found good correlations for volumetric measurements between echocardiography and cardiac CT in dogs. Indications for cardiac CT examinations in veterinary medicine include visual assessment of congenital heart anomalies, such as patent ductus arteriosus or persistent right aortic arch.12 Cardiac CT is a simple technique that can be used to rapidly assess LV function variables in patients with contraindications for cardiac MRI, such as MRI-incompatible implants. Despite these advantages over cardiac MRI, the use of cardiac CT is also limited by the need for potentially nephrotoxic contrast medium and ionized radiation for image acquisition. The great advantage of these sectional imaging techniques, compared with unidimensional and 2-D echocardiography, is the possibility for the calculation of LV function variables independent from assumptions of LV geometry.9 Changes of LV volume and EF can be used as prognostic factors for the development of heart failure (eg, chronic valvular heart disease,14 dilated cardiomyopathy,15 and LV hypertrophy).
For the evaluation of LV function variables, the Simpson method is used for cardiac CT and cardiac MRI to determine the functional variables EDV, ESV, SV, and EF The Simpson method is based on geometric shape–independent calculations and is considered accurate and reproducible.16 To our knowledge, no studies have been conducted to compare functional variables of the canine LV obtained with 3.0-T MRI and 64-slice CT via the same measurement technique. Therefore, the objective of the study reported here was to evaluate the use of cardiac MRI for the assessment of global LV function variables in dogs and compare those values with results obtained for contrast-enhanced cardiac CT examinations.
Materials and Methods
Animals—Ten healthy Beagles (7 males and 3 females) were included in the study. Mean ± SD age of the dogs was 6.5 ± 3.26 years, and mean body weight was 16.6 ± 2.08 kg. The study was approved by the Ethical Committee of the Lower Saxony State Office for Consumer Protection and Food Safety (33.9-42502-05-11A133).
On the day preceding CT and MRI examinations, a physical examination (including a hematologic analysis) was performed on each dog. In addition, an ECG was recorded and evaluated to exclude arrhythmias, and thoracic radiography was performed with the dogs positioned in right lateral recumbency for the determination of the vertebral heart scale to exclude dogs with cardiomegaly (defined as vertebral heart score > 11.2).17 To rule out hemodynamically relevant cardiovascular diseases, mitral valve regurgitation was analyzed with Doppler echocardiography. Dogs with volume-overloaded ventricles were not included in the study. Furthermore, blood pressure was measured (acceptable blood pressures were defined as SAP < 170 mm Hg and DAP < 100 mm Hg).
Study design—Dogs underwent cardiac MRI and cardiac CT on the same day during the same anesthesia episode to ensure similar hemodynamic conditions for both examinations. Cardiac MRI examination was followed by cardiac CT examination. Both examinations were conducted within a 2-hour period.
Anesthesia—Dogs were premedicated with levomethadonea (0.2 mg/kg, IV) and diazepamb (0.5 mg/kg, IV). Anesthesia was induced with propofolc administered IV (as needed to enable tracheal intubation); anesthesia was maintained after endotracheal intubation via administration of isofluraned (end-tidal volume of 1.5% in oxygen). Dogs were mechanically ventilated with a respiratory rate of 8 to 11 breaths/min during MRIe and CTf examinations. Respiratory volume was adjusted to maintain an exhaled Pco2 between 30 and 40 mm Hg, which was controlled with calibrated monitors during MRIg and CTh examinations. Heart rate was recorded via ECG. Throughout the study, saline (0.9% NaCl) solution was infused at a rate of 3 mL/kg/h. Blood pressure was measured again prior to the start of cardiac MRI and cardiac CT examinations.
Cardiac MRI—Cardiac MRI examinations were performed via a 3.0-T MRI scanneri and a personal computer with scan software.j For the cardiac MRI examinations, dogs were positioned in dorsal recumbency on the examination table. Surface coilsk were placed in an overlapping manner (2 coils on the dorsal aspect of the thorax and 2 coils on the ventral aspect of the thorax). Four MRI-compatible ECG electrodesl were applied perpendicular to each other on the left side of the thorax caudal to the elbow joint. All MRI images were acquired at end expiration at least 20 seconds after the ventilator had been turned off. Considering that acquisition of cardiac MRI images is predisposed to motion artifacts, breath-holding and ECG gating were used to avoid misalignment of acquired images.3 Although breath-holding restricts thoracic movements, ECG gating can be used to retrospectively synchronize the ECG signal with specific phases of the cardiac cycle.5 To enable us to correctly align cardiac images, survey images of the cardiac region were acquired in the transverse, coronal, and sagittal planes. An orientation line was positioned through the apex of the heart and center of the mitral valve in the survey images to enable us to adjust the right anterior oblique and 4-chamber views. To eliminate pulsing artifacts that resulted from blood flow within large vessels, a flow compensation (commonly referred to as a shim volume) was placed over the cardiac region by use of additional gradient fields. Short-axis stacks were acquired with fast-field echo sequences with the following scanning settings: echo time, 2 milliseconds; repetition time, 4 milliseconds; flip angle, 40°; matrix, 256 × 256; voxel size, 1.2 × 1.2 × 4 mm; and slice gap, 0 mm. Short-axis images were oriented perpendicular to the mitral valve and parallel to the interventricular septum (Figure 1). Short-axis images of the entire ventricle from heart base to apex were acquired during several end-expiratory inspirations. Each stack consisted of 22 slices, whereas each slice contained 30 cardiac phases. All 30 cardiac phases were measured within the same time interval, which represented at least 1 cardiac cycle.
Right anterior oblique (A) and 4-chamber views (B) used to generate a short-axis cardiac MRI image (C) of a dog. A red orientation line is positioned parallel to the mitral valve (A) and perpendicular to the septum (B) to generate a short-axis stack from the apex to the heart basis. Red arrows indicate that the views from panels A and B are used to generate the image in panel C. The midventricular short-axis image of panel C represents 1 short-axis stack with 22 heart slices, whereby each slice contains 30 cardiac phases.
Citation: American Journal of Veterinary Research 74, 7; 10.2460/ajvr.74.7.990
Cardiac MRI data analysis—All analyses were performed on a workstation with analysis software.m Evaluation of LV volume by means of the Simpson method began with selection of end-diastolic and end-systolic frames. Therefore, short-axis slices, which could be displayed in cine mode, were visually inspected. In all fast-field echo images, the cardiac lumen was bright and the myocardium was dark (Figure 1). End-diastolic images were identified as those images with maximum midventricular dilatation, primarily allocated within the first heart phase of the 30 acquired heart phases of 1 cardiac cycle. End-systolic images were determined at maximum midventricular contraction, primarily allocated within the 12th heart phase of the 30 acquired heart phases of 1 cardiac cycle. The first slice with a visible lumen was defined as the ventricular apex, whereas the most basal slice with a lumen surrounded by at least 50% myocardium (including the atrioventricular ring) was defined as the heart base. Because of LV foreshortening, ESV was primarily calculated from fewer slices than was EDV (ESV was calculated from 9 to 12 slices, whereas EDV was calculated from 10 to 14 slices). Endocardial contours were traced manually for diastolic and systolic images by drawing a line on the boundary of the cavity and myocardium (Figure 2). Papillary muscles were excluded from the ventricular cavity.
Cardiac MRI images within fast-field echo sequences for end-diastolic (A) and end-systolic (B) frames of a dog. Notice the good contrast between the blood and myocardium. End-diastolic and end-systolic frames were chosen on the basis of visual inspection of the cine mode. Notice the maximum midventricular dilatation of the left ventricle in the end-diastolic frame, whereas the maximum midventricular contraction is evident in the end-systolic frame. In both images, endocardial contours (green line) and papillary muscles (blue line) have been outlined.
Citation: American Journal of Veterinary Research 74, 7; 10.2460/ajvr.74.7.990
For the calculation of EDV and ESV, the respective short-axis areas of all slices used for the calculations were summed. Therefore, the cross-sectional area of the lumen was determined and multiplied by slice thickness and the number of slices included. The SV and EF were calculated on the basis of the EDV and ESV. Ejection fraction was calculated as (EDV – ESV)/EDV × 100. Stroke volume was calculated as EDV – ESV
Cardiac CT—Cardiac CT images were obtained with a 64-slice multidetector scanner.n Beagles were placed in dorsal recumbency on the examination table. Throughout the entire cardiac CT examination, an ECG was recorded simultaneously with periods of apnea. Therefore, ECG leadso were attached to both forelimbs and the left hind limb. Survey images were obtained first (from the thoracic inlet to the diaphragm in sagittal and dorsal orientations). For contrast enhancement, a region of interest was manually drawn in the ascending aorta of the survey image. Nonionic iodinated contrast medium (iobitridol; 2 mL/kg)p was then administered into a peripheral vein at a flow rate of 3 mL/s via a power injector.q To synchronize arrival of the bolus of contrast medium with data acquisition, a bolus-tracking technique13,18 was used. The helical scan started automatically when a threshold of 110 Hounsfield units and an additional delay time of 3.3 seconds were exceeded in the region of interest in the ascending aorta. Scan settings used for cardiac CT were as follows: detector collimation, 64 × 0.625 mm; pitch, 0.20; tube voltage, 120 kV; tube current, 400 mA; and gantry rotation time, 400 milliseconds.
Cardiac CT image reconstruction and data analysis—Data were analyzed on a workstation.r Retrospective ECG gating was applied to reconstruct images from the helical scan. For this process, raw CT data were reconstructed in 10% steps throughout the cardiac cycle. Image reconstructions from the helical to axial plane were performed with a multisegmental reconstruction algorithm.7,19 At least 10 heart phases were calculated as multiplanar reconstructions with the following settings: section thickness, 0.9 mm; increment, 0.45 mm; reconstruction matrix, 512 × 512; and field of view, 97 to 150 mm.
Images were processed with a diagnostic imaging viewers in a planar view mode. This mode represented 3 planes of the heart (horizontal long-axis plane, vertical long-axis plane, and short-axis plane), with all planes oriented perpendicular to each other (Figure 3). Automatic detection of cardiac axes required extensive manual corrections. Short-axis reformations were generated in both long-axis planes parallel to the mitral valve and perpendicular to the septum. Similar to the procedures for cardiac MRI, end-diastolic and end-systolic frames were selected by visual inspection of each of the 30 heart phases. The phase with the largest LV cavity area was defined as end diastole (most often at 0% of the cardiac cycle), and the phase with the smallest LV cavity was defined as end systole (generally at 40% of the cardiac cycle). To construct short-axis stacks, the following settings were used: number of slices, 16; slice thickness, 3.0 mm; and interslice gap, 3.7 mm (range, 3.7 to 4.1 mm). Evaluation of the LV volume was also accomplished with the Simpson method by means of analysis softwaret (Figure 4). To select slices for determination of EDV and ESV, the same conditions were applied as those used for MRI. For cardiac CT, the volume was calculated by means of semiautomated border detection. Although contours were defined on the basis of the contrast between the myocardium and lumen, they required manual correction.
Reconstructed contrast-enhanced cardiac CT images for the horizontal long-axis plane (A), short axis plane (B), and vertical long-axis plane (C) of a dog. Reconstructions are performed with the multiplanar reformation mode. The horizontal and vertical long-axis views are used for alignment of the short-axis view, which is oriented parallel to the mitral valve and perpendicular to the septum. A—Horizontal long-axis view that shows the relative positions of the short-axis (red line) and vertical long-axis (green dashed line). B—Short-axis view that shows the relative positions of the horizontal (blue line) and vertical (green dashed line) long-axis. C—Vertical long-axis view that shows the relative positions of the short axis (red line) and horizontal long axis (blue line). Tick marks on the scale in each panel are at intervals of 5 cm.
Citation: American Journal of Veterinary Research 74, 7; 10.2460/ajvr.74.7.990
Cardiac CT images of short-axis slices for the entire left ventricle of a dog from the apex (upper left image) to the heart base (lower right image). The first slice (first image in top row) was defined by a visible lumen. The last slice that was included in the volume calculation (second image from the left in bottom row) had a lumen surrounded by at least 50% myocardium. The last 2 slices in the bottom row were not used for volume calculation (crosses). In the 14 slices used for the volume calculation, the contours were manually drawn on the endocardial surface (white outline). R = Right.
Citation: American Journal of Veterinary Research 74, 7; 10.2460/ajvr.74.7.990
Statistical analysis—Graphing softwareu was used for statistical analysis and graphic representation. Heart rates and values of EDV, ESV, SV, and EF were expressed as mean and SD. The mean difference between the modalities for each of the variables was calculated with data from 3 observations. Differences were compared via the paired t test for normally distributed data or the signed rank test for nonnormally distributed data. Values of P < 0.05 were considered significant.
Linear regression analysis was used to assess the relationship between the modalities. To test the correlation between cardiac CT and cardiac MRI, the Pearson correlation coefficient was calculated for normally distributed data, and the Spearman correlation coefficient was calculated for nonnormally distributed data. Agreement between cardiac CT and cardiac MRI values was evaluated via the Bland-Altman method.20 The Bland-Altman method is based on calculation of the bias (mean difference) for each pair of values for LV EDV, ESV, SV, and EF to allow evaluation of the limits of agreement (SD around the mean difference).
Results
All dogs were in good condition. Vertebral heart scale ranged from 10 to 11.2 (mean, 10.5). Hematologic analysis revealed values within the reference range for all variables. Mean ± SD SAP was 153.3 ± 13.8 mm Hg, and mean DAP was 88.1 ± 13.5 mm Hg. During echocardiography in 4 dogs, a minimal amount of mitral valve regurgitation was evident (reflux < 20% in the left atrium), as detected with color-flow Doppler ultrasonography, and an incomplete profile in early systole was detected with continuous-wave Doppler ultrasonography.
Anesthetized dogs had a mean ± SD SAP of 112 ± 13.21 mm Hg and mean DAP of 57 ± 14.84 mm Hg. All dogs had a regular sinus rhythm with a mean heart rate of 95 ± 13.73 beats/min during cardiac CT and 95 ± 11.96 beats/min during cardiac MRI. Heart rate did not differ significantly between the modalities.
Cardiac MRI and cardiac CT examinations were conducted without complications. In all data sets for both imaging modalities, the landmarks of myocardium, LV cavity, left atrium, and LV outflow tract were clearly visible. In cardiac MRI, pulsing artifacts involving the myocardium near the apex of the heart were detected in all examinations. However, these artifacts did not disturb the identification of endocardial contours. In 7 cardiac CT data sets, blurred margins were detected, which slightly hampered the definitive identification of the endocardial contours. Despite these limitations, image quality was good.
Median and mean ± SD values from cardiac MRI and cardiac CT examinations were summarized (Table 1). Mean EDV differed significantly between CT and MRI, whereas ESV, SV, and EF did not differ significantly between the modalities. An approximate linear association between the modalities was detected by analysis of scatterplots of EDV, ESV, SV, and EF (Figure 5). Mean EDV and ESV had excellent correlation (r = 0.95) between CT and MRI values. Good correlations between modalities were found for mean SV (r = 0.67) and EF (r = 0.75; Table 2).
Scatterplots of the values for LV EDV (A), ESV (B), SV (C), and EF (D) obtained with cardiac CT (cCT) and cardiac MRI (cMRI) for 10 healthy anesthetized Beagles. Notice that there is an approximate linear association between the values for both modalities for each variable.
Citation: American Journal of Veterinary Research 74, 7; 10.2460/ajvr.74.7.990
Mean ± SD (median) values for LV function variables obtained with cardiac MRI and cardiac CT in 10 healthy anesthetized Beagles.
| Variable | Cardiac MRI | Cardiac CT |
|---|---|---|
| EDV (mL) | 39.17 ± 6.52 (38.59)a | 40.90 ± 6.93 (40.82)b |
| ESV (mL) | 20.86 ± 6.07 (19.80) | 22.09 ± 7.01 (20.92) |
| SV (mL) | 18.30 ± 2.37 (17.83) | 18.80 ± 2.36 (19.40) |
| EF (%) | 47.43 ± 7.00 (48.82) | 46.72 ± 7.52 (48.64) |
Values were determined via the Simpson method.
Values with different superscript letters differ significantly (P < 0.05; paired t test).
Correlation coefficients for values obtained with cardiac MRI and cardiac CT in 10 healthy anesthetized Beagles.
| Variable | Bias | SD | r | P value* |
|---|---|---|---|---|
| EDV (mL) | 1.74 | 2.27 | 0.95 | < 0.001 |
| ESV (mL) | 1.23 | 2.24 | 0.95 | < 0.001 |
| SV (mL) | 0.50 | 1.91 | 0.67 | 0.034 |
| EF (%) | –0.71 | 4.03 | 0.75 | 0.017 |
Values were determined via the Simpson method.
Values were considered significant at P < 0.05.
The Bland-Altman method was used to analyze the distribution of differences in EDV, ESV, SV, and EF between cardiac MRI and cardiac CT measurements (Table 2). Plotting CT values against MRI values revealed a slightly higher range of CT values for EDV, ESV, and SV, as reflected by a positive bias (Figure 6). Values for EF were an exception to this pattern because they were slightly lower for cardiac CT than for cardiac MRI, as indicated by a negative bias.
Bland-Altman plots of LV EDV (A), ESV (B), SV (C), and EF (D) values obtained with cardiac MRI (cMRI) and cardiac CT (cCT) for 10 healthy anesthetized Beagles. In each panel, the solid horizontal line represents the bias and the dotted lines represent the limits of agreement.
Citation: American Journal of Veterinary Research 74, 7; 10.2460/ajvr.74.7.990
Discussion
Echocardiography is currently accepted as the imaging modality for the assessment of LV function in veterinary medicine, whereas cardiac MRI (the criterion-referenced standard in human medicine) and cardiac CT are rarely used.3 The objectives of the study reported here were to investigate the correlation between contrast-enhanced cardiac CT and cardiac MRI measurements of LV function variables. For this purpose, 3-D data sets obtained via state-of-the-art cardiac MRI and cardiac CT devices were used to investigate LV volume in 10 Beagles. The same evaluation method was used to determine volume measurements for both modalities; therefore, similar results regarding LV function variables were expected.
We detected no significant differences between cardiac MRI and cardiac CT regarding ESV, SV, and EF. The only variable that differed significantly between the modalities was EDV, with slightly higher values for CT than for MRI. However, this variable had a linear relationship between CT and MRI, as reflected by a high correlation coefficient. Similar results were found in several human studies,6,21,22 which indicated a significant overestimation of LV volumes by cardiac CT, compared with volumes obtained with cardiac MRI. By contrast, other human studies8,18,23 revealed good agreement between cardiac CT and cardiac MRI measurements. Left ventricular volume and EF variables are commonly used to detect cardiac failure.24 Although these variables are not sufficient to describe all aspects of LV function, they are important for determining the prognosis of cardiac patients.25 As was evident in the present study, EF and LV volume variables can be effectively determined in healthy dogs by means of cardiac MRI and cardiac CT. These findings are consistent with those of a study26 in humans with LV hypertrophy, which found that cardiac MRI and tissue Doppler imaging were suitable for the measurement of diastolic function variables. However, investigators in a study27 of cats with hypertrophic cardiomyopathy found that cardiac MRI is not useful for measuring diastolic function. In comparing the LV diastolic function variables between cats with hypertrophic cardiomyopathy and clinically normal cats, no significant difference was detected between groups for cardiac MRI, whereas the diastolic variable measured with tissue Doppler imaging was significantly lower in cats with hypertrophic cardiomyopathy than in clinically normal cats.27 The insufficient cardiac MRI values in cats with hypertrophic cardiomyopathy may have been attributable to the fact that with prospective ECG gating, too few images of the heart cycle were recorded to obtain a sufficient number of images from the heart phases because of the cats’ extremely high heart rates.
Heart rate has a significant influence on LV dimension.28 An increase in heart rate results in a shorter diastolic rest period and reduces time for ventricular filling. Consequently, EDV decreases, whereas ESV remains at the same value because the duration of systole remains nearly unchanged at higher heart rates.29 In the present study, heart rates did not differ significantly during cardiac CT and cardiac MRI examinations; thus, it can be concluded that heart rate exerted a similar influence on the LV volume measurements for MRI and CT.
Both volumetric imaging techniques have excellent spatial resolution,30 whereas cardiac CT is limited in its temporal resolution. The use of subsecond gantry rotation times and multisegmental reconstruction algorithms has remarkably improved the temporal resolution for 64-slice cardiac CT over that of previous CT machines.30 However, the temporal resolution of cardiac CT is still not equal to that of cardiac MRI. To avoid discontinuities of heart images during cardiac CT, a low pitch is required at fast gantry rotation times.31 Further improvements to temporal resolution are made by the use of a multisegment reconstruction algorithm, which adds subvolumes from consecutive cardiac cycles by means of ECG gating to generate a complete cardiac data set.19,32 Therefore, higher heart rates are beneficial because at higher heart rates more R-R intervals are used for image reconstruction and, consequently, the temporal resolution is improved.29
Optimization of the temporal resolution for cardiac MRI can also be achieved by reconstructing a single image from multiple heart cycles.33 Therefore, raw data are acquired by encoding lines of k-space. To encode a sufficient number of lines, the process must be repeated several times for as many cardiac cycles as necessary to reproduce an entire volume set5; thus, all reconstructed images are an average of several heartbeats. The drawback of this technique in cardiac MRI and cardiac CT is that there can be incorrect data reconstruction because of irregularities of the cardiac rhythm, such as sinus arrhythmia or extra systoles.13 Neither arrhythmias nor extra systoles were observed in the present study, so good reconstruction of the data was assumed. In several studies,7,34,35 investigators have highlighted the importance of high temporal resolution for precise qualitative and quantitative assessment of LV function. Moreover, the lower temporal resolution of cardiac CT has been described as a potential source of error.36,37 During the relatively short period of late systole, ESV is prone to error in cases of low temporal resolution, which consequently results in overestimation of ESV by cardiac CT. However, in the study reported here, only EDV values were significantly higher for CT than for MRI. Comparable results were reported by investigators of another study,6 who described an overestimation of EDV and ESV for a study with a design similar to that of the present study. These facts imply that limited temporal resolution for cardiac CT was not the reason for the difference in EDV values between CT and MRI in the present study. The difference may have been caused by the rapid injection of a relatively large volume of contrast medium immediately before the CT examination, which could have resulted in a temporary increase in preload and negative inotropic effects.6
Another reason for discrepancies between cardiac MRI and cardiac CT measurements in the present study could have been the selection procedure used to identify the most basal slice of the short-axis stacks.38 Despite the precise definition of the basal slice (ie, lumen surrounded by at least 50% myocardium), deciding whether a slice was still part of the LV volume was difficult in a few analyses. Consequently, erroneous inclusion or exclusion of the basal section within the LV volume can result in substantial overestimation or underestimation of volume.13 In slices from other locations of the heart, inaccuracies in tracing of the endocardial border only cause small errors in volume measurement.13 Moreover, in the case of oblique short-axis slices, volume can be overestimated because of inclusion of portions of the left atrium or aorta.9 Nevertheless, the calculation of LV volume from short-axis views by means of the Simpson method is considered the most accurate and reproducible method for determination of LV volume.16,33
Another potential source of error was the use of different slice thicknesses for cardiac MRI and cardiac CT examinations. For CT, a slice thickness of 3 mm with an interslice gap of 4 mm was used, whereas MRI data were assessed with a slice thickness of 4 mm without an interslice gap. In general, thinner slices are more accurate.39 However, for MRI, the use of thinner slices would have prolonged the anesthetic episode; therefore, a slice thickness of 4 mm was used as a compromise between accuracy and anesthesia duration. In cardiac CT, thicker slices result in blurring artifacts of endocardial margins, which can result in a slight overestimation of volume.13 In the present study, we also found blurring artifacts with thinner slices. However, in a previous study,39 investigators evaluated the importance of slice thickness on accuracy of volume determination and found no significant differences in the calculated volume between slices with a thickness of 2 or 5 mm.
Data acquisition during cardiac MRI can also be influenced by being performed over several periods of end inspiration. Repositioning of the heart after breathing may cause erroneous slice position and can result in and ultimately lead to inaccurate volume measurements. To eliminate this source of error, the scans were always started during end expiration because examinations in this respiration phase have the best reproducibility.33 In comparison to cardiac MRI, cardiac CT appears to be less prone to repositioning errors of the heart because data collection takes place during a single end expiration.30 Both imaging modalities are noninvasive and also exclude papillary muscles from the LV cavity. However, the method used to exclude the papillary muscles differs between MRI and CT, which could have been a reason for the small measurement differences in LV function variables between the imaging modalities in the present study. Nevertheless, exclusion of the papillary muscles during CT and MRI results in a more accurate volume measurement, compared with that of echocardiography, for which papillary muscles are included in the lumen because of their poorer demarcation from the cavity. Moreover, enlarged or deformed hearts can be examined more accurately with CT and MRI than with echocardiography because the calculation of LV function variables within unidimensional and 2-D echocardiography is based on assumptions of LV geometry.24 However, we do not currently expect that veterinarians will routinely use cardiac CT and cardiac MRI in veterinary cardiology because of the need for general anesthesia of patients. Nevertheless, CT and MRI are both effective for the determination of LV volume and have their own advantages and disadvantages.
The advantages of cardiac CT for the evaluation of cardiac function variables include a short data acquisition time and the ability to repeat slice reformations. A short examination time is especially useful to reduce the duration of anesthesia for dogs with severe heart failure. The overlapping technique of the spiral CT examination enables a fast and complete record of heart action. A drawback of this technique is the higher radiation exposure, compared with that for sequential examination.22 An additional benefit of data acquisition with cardiac CT is the possibility of renewal of short-axis images as many times as is needed to achieve an adequate orientation, given that correctly delineated axes are necessary for precise volume measurements.
The benefits of cardiac MRI include excellent spatial and temporal resolution that results in high image quality without exposing the patient to x-ray irradiation or iodine-based contrast agents. However, cardiac MRI is contraindicated for patients with ferrous implants or severe heart failure because a prolonged duration of anesthesia is necessary.
The present study had some limitations. One was the need for anesthesia, which contributes to hemodynamic changes (eg, decreased heart rate or decreased myocardial contractility that can result in impaired function).7 Further limitations were the small number of dogs in the study and homogeneity for the population of 10 Beagles, which hindered extrapolation of the results to larger dog populations or other breeds. Four of the 10 dogs had minimal mitral valve regurgitation, but these valve insufficiencies were extremely small and not hemodynamically relevant because values for the LV chambers were within reference limits, as proven with echocardiographic examination.
Both cardiac MRI and cardiac CT achieved similar results regarding LV volume and function variables. Adequate spatial and temporal resolution leads to good image quality for cardiac MRI and cardiac CT data sets and enables reliable assessment of LV volume. Cardiac MRI and cardiac CT have different inclusion and exclusion criteria; therefore, they can be used in a complementary manner. Although their use is limited in veterinary practice, they are superior in accuracy and reproducibility, compared with conventional echocardiography. Thus, earlier detection of impaired cardiac function can be expected.
ABBREVIATIONS
| DAP | Diastolic arterial blood pressure |
| EDV | End-diastolic volume |
| EF | Ejection fraction |
| ESV | End-systolic volume |
| LV | Left ventricular |
| SAP | Systolic arterial blood pressure |
| SV | Stroke volume |
L-Polamivet, Intervet Deutschland GmbH, Unterschleissheim, Germany.
Diazepam-ratiopharm, 10 mg/2 mL, Ratiopharm GmbH, Ulm, Germany.
Narcofol, 10 mg/mL, CP-Pharma Handelsgesellschaft mbH, Burgdorf, Germany.
Isofluran CP, CP-Pharma Handelsgesellschaft mbH, Burgdorf, Germany.
Ventilog C, Draeger Medical AG, Lübeck, Germany
Ventilog 2, Draeger Medical AG, Lübeck, Germany.
PM8050 MRI, Draeger Medical AG, Lübeck, Germany
Infinity Delta, Draeger Medical AG, Lübeck, Germany
Achieva, 3.0 T, Philips Medical Systems, Best, The Netherlands.
MR Systems Achieva, Philips Medical Systems, Best, The Netherlands.
Sense Flex small/medium, Philips Medical Systems, Best, The Netherlands.
Radiotranslucent foam monitoring electrodes, Boeblingen, Germany.
Extended MR Workspace, Philips Medical Systems, Best, The Netherlands.
Brilliance 64, Philips Medical Systems, Cleveland, Ohio.
SilverTrace ECG electrodes, GE Medical Systems, Freiburg, Germany.
Xenetix, 350 mg of I/mL, Guerbet GmbH, Sulzbach, Germany.
Medrad Vistron CT injection system, Medrad, Warrendale, Pa.
Extended Brilliance Workspace, Philips Medical Systems, Best, The Netherlands.
Cardiac viewer, Philips Medical Systems, Best, The Netherlands.
LV/RV Analysis, Philips Medical Systems, Best, The Netherlands.
GraphPad Prism, version 5.0 for Windows, GraphPad Software Inc, San Diego, Calif.
References
1. McMurray JJ, Adamopoulos S, Anker SD, et al. ESC guidelines for the diagnosis and treatment of acute and chronic heart failure 2012: The Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2012 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association (HFA) of the ESC. Eur Heart J 2012; 33: 1787–1847.
2. Kjaergaard J, Petersen CL, Kjaer A, et al. Evaluation of right ventricular volume and function by 2D and 3D echocardiography compared to MRI. Eur J Echocardiogr 2006; 7: 430–438.
3. Gilbert SH, McConnell FJ, Holden AV, et al. The potential role of MRI in veterinary clinical cardiology Vet J 2010; 183: 124–134.
4. Contreras S, Vazquez JM, Miguel AD, et al. Magnetic resonance angiography of the normal canine heart and associated blood vessels. Vet J 2008; 178: 130–132.
5. Mai W, Weisse C, Sleeper MM. Cardiac magnetic resonance imaging in normal dogs and two dogs with heart base tumor. Vet Radiol Ultrasound 2010; 51: 428–435.
6. Sugeng L, Mor-Avi V, Weinert L, et al. Quantitative assessment of left ventricular size and function: side-by-side comparison of real-time three-dimensional echocardiography and computed tomography with magnetic resonance reference. Circulation 2006; 114: 654–661.
7. Mahnken AH, Spuentrup E, Niethammer M, et al. Quantitative and qualitative assessment of left ventricular volume with ECG-gated multislice spiral CT: value of different image reconstruction algorithms in comparison to MRI. Acta Radiol 2003; 44: 604–611.
8. Greupner J, Zimmermann E, Grohmann A, et al. Head-to-head comparison of left ventricular function assessment with 64-row computed tomography, biplane left cineventriculography, and both 2- and 3-dimensional transthoracic echocardiography: comparison with magnetic resonance imaging as the reference standard [Erratum published in J Am Coll Cardiol 2012; 60:481]. J Am Coll Cardiol 2012; 59: 1897–1907.
9. Dell'Italia LJ, Blackwell GG, Pearce DJ, et al. Assessment of ventricular volumes using cine magnetic resonance in the intact dog. A comparison of measurement methods. Invest Radiol 1994; 29: 162–167.
10. Markiewicz W, Sechtem U, Kirby R, et al. Measurement of ventricular volumes in the dog by nuclear magnetic resonance imaging. J Am Coll Cardiol 1987; 10: 170–177.
11. Hockings PD, Busza AL, Byrne J, et al. Validation of MRI measurement of cardiac output in the dog: the effects of dobutamine and minoxidil. Toxicol Mech Methods 2003; 13: 39–43.
12. Henjes CR, Nolte I, Wefstaedt P. Multidetector-row computed tomography of thoracic aortic anomalies in dogs and cats: patent ductus arteriosus and vascular rings. BMC Vet Res 2011; 7:57.
13. Henjes CR, Hungerbuehler S, Bojarski IB, et al. Comparison of multi-detector row computed tomography with echocardiography for assessment of left ventricular function in healthy dogs. Am J Vet Res 2012; 73: 393–403.
14. Ljungvall I, Hoglund K, Carnabuci C, et al. Assessment of global and regional left ventricular volume and shape by real-time 3-dimensional echocardiography in dogs with myxomatous mitral valve disease. J Vet Intern Med 2011; 25: 1036–1043.
15. Tidholm A, Haeggstrom J, Borgarelli M, et al. Canine idiopathic dilated cardiomyopathy. Part I: aetiology, clinical characteristics, epidemiology and pathology. Vet J 2001; 162: 92–107.
16. Hergan K, Schuster A, Frühwald J, et al. Comparison of left and right ventricular volume measurement using the Simpson's method and the area length method. Eur J Radiol 2008; 65: 270–278.
17. Kraetschmer S, Ludwig K, Meneses F, et al. Vertebral heart scale in the Beagle dog. J Small Anim Pract 2008; 49: 240–243.
18. Maffei E, Messalli G, Martini C, et al. Left and right ventricle assessment with cardiac CT: validation study vs. cardiac MR. Eur Radiol 2012; 22: 1041–1049.
19. Boese JM, Bahner ML, Albers J, et al. Optimizing temporal resolution in CT with retrospective ECG gating [in German]. Radiologe 2000; 40: 123–129.
20. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–310.
21. Brodoefel H, Kramer U, Reimann A, et al. Dual-source CT with improved temporal resolution in assessment of left ventricular function: a pilot study. AJR Am J Roentgenol 2007; 189: 1064–1070.
22. Heuschmid M, Rothfuss J, Schroder S, et al. Left ventricular functional parameters: comparison of 16-slice spiral CT with MRI [in German]. Rofo 2005; 177: 60–66.
23. Mahnken AH, Spuntrup E, Wildberger JE, et al. Quantification of cardiac function with multislice spiral CT using retrospective EKG-gating: comparison with MRI [in German]. Rofo 2003; 175: 83–88.
24. Wess G, Maurer J, Simak J, et al. Use of Simpson's method of disc to detect early echocardiographic changes in Doberman Pinschers with dilated cardiomyopathy. J Vet Intern Med 2010; 24: 1069–1076.
25. Boon JA. Dilated cardiomyopathy. In: Boon JA, ed. Veterinary echocardiography. Ames, Iowa: Wiley-Blackwell, 2011;381–396.
26. Kudelka AM, Turner DA, Liebson PR, et al. Comparison of cine magnetic resonance imaging and Doppler echocardiography for evaluation of left ventricular diastolic function. Am J Cardiol 1997; 80: 384–386.
27. MacDonald KA, Kittleson MD, Garcia-Nolen T, et al. Tissue Doppler imaging and gradient echo cardiac magnetic resonance imaging in normal cats and cats with hypertrophic cardiomyopathy. J Vet Intern Med 2006; 20: 627–634.
28. Jacobs G, Mahjoob K. Influence of alterations in heart rate on echocardiographic measurements in the dog. Am J Vet Res 1988; 49: 548–552.
29. Mahnken AH, Hohl C, Suess C, et al. Influence of heart rate and temporal resolution on left-ventricular volumes in cardiac multislice spiral computed tomography: a phantom study. Invest Radiol 2006; 41: 429–435.
30. Juergens KU, Grude M, Maintz D, et al. Multi-detector row CT of left ventricular function with dedicated analysis software versus MR imaging: initial experience. Radiology 2004; 230: 403–410.
31. Primak AN, McCollough CH, Bruesewitz MR, et al. Relationship between noise, dose, and pitch in cardiac multi-detector row CT. Radiographics 2006; 26: 1785–1794.
32. Ohnesorge B, Flohr T, Becker C, et al. Cardiac imaging with rapid, retrospective ECG synchronized multilevel spiral CT [in German]. Radiologe 2000; 40: 111–117.
33. Mahnken AH, Gunther RW, Krombach GA. The basics of left ventricular functional analysis with MRI and MSCT [in German]. Rofo 2004; 176: 1365–1379.
34. Juergens KU, Maintz D, Grude M, et al. Multi-detector row computed tomography of the heart: does a multi-segment reconstruction algorithm improve left ventricular volume measurements? Eur Radiol 2005; 15: 111–117.
35. Miller S, Simonetti OP, Carr J, et al. MR imaging of the heart with cine true fast imaging with steady-state precession: influence of spatial and temporal resolutions on left ventricular functional parameters. Radiology 2002; 223: 263–269.
36. Koch K, Oellig F, Oberholzer K, et al. Assessment of right ventricular function by 16-detector-row CT: comparison with magnetic resonance imaging. Eur Radiol 2005; 15: 312–318.
37. Takx RA, Moscariello A, Schoepf UJ, et al. Quantification of left and right ventricular function and myocardial mass: comparison of low-radiation dose 2nd generation dual-source CT and cardiac MRI. Eur J Radiol 2012; 81: e598–e604.
38. Schalla S, Nagel E, Lehmkuhl H, et al. Comparison of magnetic resonance real-time imaging of left ventricular function with conventional magnetic resonance imaging and echocardiography. Am J Cardiol 2001; 87: 95–99.
39. Cui W, Kondo T, Anno H, et al. The accuracy and optimal slice thickness of multislice helical computed tomography for right and left ventricular volume measurement. Chin Med J (Engl) 2004; 117: 1283–1287.
