Nuclear Medicine Imaging of Ventricular Function

Published on 13/02/2015 by admin

Last modified 22/04/2025

Print this page

This article have been viewed 2048 times

CHAPTER 56 Nuclear Medicine Imaging of Ventricular Function

Left ventricular (LV) function is an important, well-established indicator of patient prognosis. Radionuclide imaging techniques are established as the most highly reproducible methods to assess LV ejection fraction noninvasively. Comprehensive textbooks have summarized the results of studies over several decades showing the high reproducibility and accuracy of gated equilibrium blood pool scintigraphy and radionuclide angiography in determining LV and right ventricular (RV) ejection fraction measurements.¹^–⁴ Cardiac mortality is strongly associated with impaired ventricular function as assessed by decreased LV ejection fraction. Additionally, identifying dysfunctional segments in viable myocardium is crucial in clinical management. Numerous studies have shown an almost threefold higher incidence of cardiac mortality in medically treated patients with viable compared with nonviable myocardium.⁵ The identification of focal dysfunctional segments is highly relevant to clinical management.

The introduction of Tc-99m radiotracers has made gated myocardial perfusion imaging (MPI) readily available, easily feasible, and now a standard technique requiring little additional effort or cost.⁶ Because MPI is the most commonly performed procedure in nuclear cardiology, evaluation of LV function in nuclear cardiology is most often performed in conjunction with MPI. Similarly, assessment of ventricular function by gated MPI can be performed with positron emission tomography (PET).⁷

Gated equilibrium blood pool imaging is a well-established technique commonly used to assess ventricular function quantitatively in oncology patients receiving cardiotoxic chemotherapy. High reproducibility, technical simplicity, and availability, make this study the procedure of choice in evaluating LV function sequentially. Although first-pass radionuclide angiocardiography is a well-established and elegant technique to determine ejection fraction, equilibrium blood pool imaging provides the ability to view ventricular function from multiple planar projections, or in a fully three-dimensional manner using single photon emission computed tomography (SPECT).

For quantitative, functional analysis of the right ventricle by RV ejection fraction, first-pass angiocardiography currently remains the method of choice because planar equilibrium blood pool imaging cannot separate underlying or adjacent blood pool structures sufficiently for an accurate determination of RV ejection fraction. Newer techniques, such as tomographic ECG gated blood pool (GBP) imaging, are becoming more widespread as quantitative software programs are being developed and validated. These newer techniques have the potential to evaluate simultaneously and quantitatively RV and LV size and function, with high spatial and temporal resolution. Elegant first-pass techniques have been developed to evaluate intraventricular shunts quantitatively; the need for these is relatively uncommon, and they are described elsewhere.¹

This chapter describes the most commonly performed radionuclide techniques to assess ventricular function. First, gated SPECT MPI is discussed. These principles are generally applicable to gated PET assessment of ventricular function. As PET MPI becomes more prevalent in the future, this technique is expected to become increasingly important. Next, equilibrium GBP imaging is reviewed briefly. Finally, RV functional evaluation with first-pass radionuclide angiocardiography is discussed.

LEFT VENTRICULAR FUNCTIONAL ASSESSMENT FROM GATED MYOCARDIAL PERFUSION IMAGING

ECG gating to assess ventricular function is now a standard, integral part of MPI. The quantitative evaluation of LV ejection fraction has exhibited high reproducibility and accuracy. Compared with CT, LV ejection fraction assessed by SPECT is not significantly different, although absolute values for end-diastolic volume and end-systolic volume are different with SPECT.⁸ As expected, LV functional impairment quantified by SPECT is associated with a poor prognosis, including cardiac death. LV ejection fraction less than 45% and LV end-systolic volume greater than 70 mL were associated with approximately 8% to 9%/yr cardiac death rate compared with approximately 1%/yr for LV ejection fraction 45% or greater and end-systolic volume 70 mL or less.⁹ Although perfusion defect reversibility is most predictive of future myocardial infarction, LV ejection fraction is the strongest predictor of mortality.¹⁰

Several studies have shown the clinical utility with improved specificity by identifying normal wall motion and wall thickening in areas of decreased activity because of tissue attenuation. This improvement in specificity may be particularly important in women because of breast attenuation artifacts. The gating information also assists in improving the degree of certainty when interpreting studies as definitely normal or definitely abnormal.⁶^,¹¹ In a prospective study of 115 women, gated Tc 99m sestamibi significantly improved specificity (92%) compared with the same study without gating (84%; P = .0004) or with thallium 201 without gating (59%).¹²

In addition, gating improves sensitivity for detection of coronary artery disease (CAD) in the subset of patients with multivessel disease (“balanced disease”) who have transient, stress-induced wall motion abnormalities that are not detected by MPI alone.¹³ Additional wall motion information significantly increased the sensitivity for detection of multivessel CAD without adversely affecting specificity.¹⁴ In this study, the sensitivity for multivessel disease was significantly increased from 46% to 60%, and sensitivity for three-vessel disease significantly increased from 10% to 25%.¹⁴ Principles of gated MPI with SPECT are similar to gating with PET radiotracers such as ammonia N 13 or rubidium 82, providing regional and global wall motion information.¹⁵^,¹⁶ The more recent development of combined PET/CT and SPECT/CT instrumentation may also permit additional, independent, high spatial resolution CT assessment of regional myocardial function.¹⁷

Electrocardiogram Gated Acquisition

Analogous to traditional methods of acquiring planar gated studies, the ECG is used to bin counts from the various phases of the cardiac cycle temporally during data acquisition. Because of count limitations, most studies currently use eight time frames per cardiac cycle. In patients with irregular rhythms, such as atrial fibrillation, ventricular function may be better assessed with methods that do not rely on data from multiple, averaged heartbeats (e.g., echocardiography). If gated MPI is used, the averaged data may be subject to inaccuracies because of misplacement of counts within the cardiac cycle, and careful inspection of the ECG and gating data is necessary to ensure that appropriate beats with similar RR intervals are included for analysis.¹⁸^,¹⁹

Data Processing, Reconstruction, and Analysis

In most cases, appropriate binning of the data into the various phases of the cardiac cycle results in accurate representation of LV function. Tomographic images are typically reconstructed with filtered back-projection, and sophisticated edge detection programs are used to define ventricular boundaries. From these volumes, time-activity curves are generated, indices of ventricular function are derived, and a cine display is generated for visual assessment of regional and global LV function.

Well-developed programs to evaluate LV wall motion have been commercially available for several years.²⁰^–²² Generally, these use boundary detection algorithms, count densities, or a combination of variables to define the endocardial and epicardial borders for LV volume calculations. Another important parameter is LV wall thickening. Detected activity within the myocardium depends on radioactive concentration and volume (or thickness) of the myocardium. When small objects (LV myocardial wall thickness) are below twice the full-width half-maximum of the system resolution, detected activity within the ventricular wall appears lower than the true activity because of partial volume effects.²³ Because myocardial wall thickness is susceptible to the partial volume effect, a higher wall thickness at end-systole corresponds to higher detected activity by SPECT. The degree of regional wall thickening is related to the change in detected regional myocardial wall activity. Detected activity at end-diastole typically is lower than that at end-systole because of differences in ventricular wall thickness. Some algorithms assume there is an exact linear relationship between myocardial wall thickening and percent systolic count increases.

Cardiac volume and myocardial thickening can be appreciated visually for cine display formats of individual tomographic slices and computer-generated derived surface renderings. In addition, quantitative, physiologic indices are derived from tomographic volumes and from counts within the myocardium. Fourier analysis is applied to the time-activity curve of each data element in the volume (voxel) to generate parametric images of contraction. Fourier analysis forms the basis of phase and amplitude analysis by which the pattern of myocardial contraction and regional wall motion is assessed quantitatively. More recently, high correlation of LV ejection fraction measurements was reported between techniques in hybrid instrumentation combining high-resolution 64-slice CT with SPECT; mean values were not significantly different.⁸

As described earlier, the primary potential benefits of concurrent wall motion assessment include (1) improved specificity in cases of fixed regions of decreased activity (improving detection of attenuation artifacts), and (2) improved sensitivity in detection of multivessel CAD or three-vessel disease (i.e., detection of abnormal wall motion in patients with “balanced disease”). With respect to increased specificity, this may be particularly beneficial in women with severe anterior breast attenuation, and in men with severe inferior diaphragmatic attenuation. The presence of completely normal wall motion and wall thickening is strong evidence that large transmural infarction has not occurred. Although small subendocardial infarction may not be detected with gated perfusion imaging, normal wall motion and normal LV function are good prognostic indicators, which can greatly aid in the assessment of prognosis.

With respect to sensitivity, exercise-induced LV dysfunction has shown improvement in sensitivity for detection of multivessel CAD over perfusion information alone; it also provides further information regarding the clinical significance of known disease. This information may be clinically important in patients with multivessel CAD, who are among the patients most likely to benefit from further therapy.

PITFALLS AND SOLUTIONS

Potential limitations of this technique should also be mentioned. In patients with prior infarction, boundaries in regions of low or absent tracer uptake may be difficult to define accurately. Although edge detection algorithms are in most cases beneficial, in some cases the extremely low activity renders myocardial boundary definition very difficult and susceptible to computer boundary definition errors. In a canine experimental model, boundary detection and reproducibility were adversely affected by prior infarction, low tracer dose administration, and high background activity.²⁴ Additionally, in patients with small LV cavities (e.g., small women), the endocardial borders may be difficult to identify, and an underestimation of LV cavity size at end systole may lead to artifactually high LV ejection fraction values.⁶

As with all ECG gated techniques, arrhythmias may also interfere with appropriate data collection, and ECG data should be inspected carefully. Differences in boundary detection algorithms exist in different software packages, and volumes and LV ejection fraction values may differ from one algorithm to the next because of variability in data analysis methods. Comparing ejection fraction or volume calculations produced by more than one nuclear cardiology software package, or comparing against values obtained from another imaging modality such as MRI, should be performed carefully and should take into account methodologic differences, including the normal limits specific to each algorithm.²⁵

LV function assessed with MPI is a well-established method that has high prognostic significance. Gated MPI provides functional information highly predictive of cardiac mortality; this is independent of MPI data, which is also predictive of future cardiac events, including myocardial infarction. Gating improves the sensitivity in the detection of multivessel CAD and increases specificity or normalcy rate in patients with attenuation artifacts, such as anterior breast attenuation in women and LV inferior wall attenuation in men.

TECHNIQUES

Equilibrium Gated Blood Pool Imaging

Planar GBP imaging with Tc 99m–labeled red blood cells (RBCs) is a well-established and validated technique with the highest reproducibility for LV ejection fraction among all noninvasive imaging procedures. Several factors contribute to this high reproducibility: (1) Tc-99m RBC labeling efficiency is high, (2) LV ejection fraction is calculated by counts within boundaries that are defined by automated detection algorithms, and (3) high count images can be obtained within a short imaging time.

Indications

This technique has been well established as the clinical method of choice to assess objectively oncology patients at risk of developing cardiotoxicity from chemotherapy.²⁶ Although echocardiography is rapid and results in little patient discomfort, its low quantitative reproducibility obviates its use in standard chemotherapeutic monitoring. Radionuclide gated equilibrium blood pool imaging similarly provides rapid assessment with little discomfort, but the high quantitative reproducibility, technical simplicity, and widespread availability, combined with relatively low cost, make this the method of choice for quantitative assessment of LV function in this population.

Technique Description

All three methods of radiolabeling RBCs with Tc-99m pertechnetate (in vivo, modified in vivo, and in vitro) result in high radiolabeling efficiency that is typically approximately 85% to 90% or greater. Although the in vivo method typically produces the lowest (>85%) radiolabeling efficiency, it is the simplest and provides excellent contrast sufficient for reproducible LV ejection fraction measurements in most cases. For the in vivo method, stannous pyrophosphate is first administered intravenously to reduce the hemoglobin within the RBCs. Approximately 10 to 20 minutes afterward, Tc-99m pertechnetate is administered intravenously, effectively binding to the reduced hemoglobin.

If a quantitative measurement of RV ejection fraction is desired, the radiotracer can be administered with the gamma camera in the right anterior oblique projection using a “gated list mode” acquisition. This acquisition acquires and stores the precise location of all detected photons and when they occurred with respect to the ECG trigger throughout the acquisition. After acquisition, these data are retrospectively processed by rebinning the data into the appropriate phases of the ECG cycle. Typically, the data obtained during radiotracer bolus transit through the right ventricle occurs in approximately three to five heartbeats (or ECG cycles). The rebinned data from these heartbeats are combined to obtain sufficient counts for quantitative analysis. Approximately 20 to 30 minutes afterward, planar equilibrium blood pool images are obtained in the anterior, best septal separation (left anterior oblique 45 degrees [LAO 45]), and left lateral positions to evaluate ventricular wall motion and LV ejection fraction.

The anterior view evaluates anterior and apical segments. The best septal separation view (typically LAO 45) displays septal, inferoapical, and lateral walls. The left lateral view shows anterior, inferior, and apical regions. The best septal separation (LAO 45) has the least amount of overlapping blood pool from adjacent structures. This view is used to calculate LV ejection fraction according to the following formula:

for background-corrected end-diastolic and end-systolic counts, where LVED is LV end-diastolic volume counts and LVES is LV end-systolic volume counts after background correction.

Background counts typically are determined using an extraventricular region defined from the lateral to inferolateral LV wall. The normal LV ejection fraction is defined as 50% or greater. Reproducibility data show that greater than 5% difference in LV ejection fraction constitutes a statistically significant difference. Curve fitting with Fourier analysis is performed for each voxel in the best septal separation view (typically LAO 45); this provides parametric phase and amplitude images representing the onset and magnitude of contraction. These aid in image processing by defining ventricular boundaries and valve planes; they also aid in image interpretation by providing an objective method to assess the pattern of contraction and regional asynchrony (Figs. 56-1 and 56-2).

FIGURE 56-1 A, RV and LV wall motion are normal in a 58-year-old woman with endometrial carcinoma undergoing imaging before chemotherapy. LV ejection fraction (LVEF) = 65%. End-diastolic (top row) and end-systolic (bottom row) images show normal ventricular sizes and normal end-systolic volumes. B, Parametric amplitude image (left image) of the planar LAO 45 acquisition shows high and relatively uniform change in counts in RV and LV regions. The parametric phase image (middle image) shows relatively uniform peak activity in the ventricles, indicating a uniform contraction pattern. The histogram of phase shows a narrow and uniform ventricular phase; this is out of phase compared with the atrial voxels.

FIGURE 56-2 A, Severe RV and LV global hypokinesis in 55-year-old man with obesity and severe CAD who is also status post myocardial infarctions, coronary artery bypass graft surgeries, and multiple stents. The left ventricle and right ventricle are dilated, and their respective end-systolic volumes are abnormally increased. Images appear “noisy” because of scatter from soft tissue attenuation. LV ejection fraction (LVEF) = 16%. End-diastolic (top row) and end-systolic (bottom row) images show markedly enlarged ventricular cavities and abnormally high end-systolic volumes. B, Parametric amplitude image (left image) of the planar LAO 45 acquisition shows low ventricular change in counts for RV and LV regions. Best region of LV contraction is the lateral wall (arrow); most of the LV cavity shows near-absent change in counts. The parametric phase image (middle image) shows relatively nonuniform peak activity in the ventricles, indicating a nonuniform contraction pattern. The histogram of phase shows a wide and nonuniform ventricular phase; this is out of phase compared with the atrial voxels.

RV wall motion can be assessed visually (Fig. 56-3); however, planar images do not permit accurate or reproducible RV ejection fraction calculations because of the adjacent structures that overlap the right ventricle. This issue can be addressed by calculating RV ejection fraction from a gated first-pass list mode radionuclide angiogram or by tomographic methods, as described later. Specific pathologic entities (e.g., atrial septal defect, ventricular septal defect) can also affect ventricular size and function as seen on GBP imaging and have been described elsewhere in detail.¹

FIGURE 56-3 A, RV enlargement and hypokinesis; LV size and wall motion are normal. The RV enlargement rotates the best septal separation view to left anterior oblique 70 degrees (LAO 70). LV ejection fraction (LVEF) = 74%. End-diastolic (top row) and end-systolic (bottom row) images show markedly enlarged RV cavity at end-diastole and end-systole. LV cavity is normal in size at end-diastole and end-systole. B, The parametric amplitude image (left image) of the planar LAO 70 acquisition shows relatively lower ventricular change in counts for the right ventricle. The LV change in counts is high, relatively uniform, and normal. The parametric phase image (middle image) shows relatively uniform peak activity in the left ventricle, indicating a uniform contraction pattern. The right ventricle is also uniform, but is slightly delayed compared with the left ventricle. The histogram of phase shows a wide ventricular phase with the right ventricle slightly delayed compared with the left ventricle; these are both out of phase compared with the atrial voxels.

Tomographic blood pool imaging has shown high correlation of RV ejection fraction with planar first-pass techniques.²⁷ More recently, automated analysis software for tomographic GBP imaging has confirmed high accuracy and reproducibility in the assessment of RV function as confirmed by MRI and echocardiography.²¹^,²⁸ Because of the increasing incidence of heart failure, accurate biventricular functional assessment may become increasingly important in clinical practice and investigations.

Advantages of tomography include higher contrast, three-dimensional localization, potential for higher accuracy with respect to ventricular volume and wall motion measurements, and simultaneous assessment of the right and left ventricles. These advantages are particularly useful in assessing the right ventricle because of its irregular shape, which is difficult to model using simple assumptions (Fig. 56-4). It has been shown more recently that LV and RV measurements of function are highly reproducible by tomographic blood pool techniques.²⁹ Consequently, interest has re-emerged in the use of GBP SPECT to identify candidates with ventricular asynchrony who may benefit from cardiac resynchronization therapy. GBP SPECT incorporates analysis of RV and LV contraction phase information, in contrast to gated MPI phase analysis, which currently measures only contraction phases for the left ventricle, but not for the right ventricle.³⁰

FIGURE 56-4 A, Markedly abnormal GBP SPECT of a 75-year-old man with heart failure. Boundaries outline RV and LV regions for analysis. The left ventricle and right ventricle are prominent in size. Short-axis from apex to base (rows 1 to 3); vertical long-axis from right ventricle to left ventricle (rows 4 and 5); and horizontal long-axis from inferior to anterior (rows 6 to 8). B, Quantitative analysis of GBP SPECT. Analysis shows LV apical dyskinesis, which was confirmed by MRI (not shown). LV ejection fraction by GBP SPECT is 21%, and LV ejection fraction by MRI was 23%. Right ventricle (top row) and left ventricle (bottom row) regional analyses. Phase histogram (left column) and polar plots (right column) of regional segmental analysis. Phase histogram (left column) shows multiple, wide peaks for right ventricle and left ventricle indicating abnormal contraction. A single narrow peak would indicate normal contraction. Polar plots (right column) show regional measurements of phase with a wide variation indicating asynchronous contraction. ES, end-systolic.

(Courtesy of K. Nichols, PhD.)

First-Pass Radionuclide Angiography

Clinical Indications

The current primary clinical application for first-pass angiography is to evaluate quantitatively RV function. As briefly mentioned earlier, RV ejection fraction calculations cannot be accurately performed using planar gated equilibrium blood pool techniques. First-pass techniques are elegant, quantitative methods to assess intraventricular shunts. In the past, first-pass techniques used in conjunction with exercise protocols were used to diagnose CAD noninvasively by detection of stress-induced wall motion abnormalities. Generally, first-pass techniques provide relatively lower counts, however, and are more technically challenging to perform compared with tomographic GBP techniques. They are much less commonly used in clinical practice because of the relatively uncommon prevalence specifically requiring their use, and the emergence of alternative techniques, such as contrast echocardiography, which may provide sufficient information for clinical management.

Technical Aspects

Data Acquisition

The technical details of acquisition are extensive and have been reviewed more recently.⁵ To perform planar gated first-pass imaging, an ECG gated list mode acquisition is used in conjunction with Tc-99m pertechnetate injection of 740 to 1110 MBq (20 to 30 mCi). Because this is the same radionuclide activity used for GBP imaging, this study can be combined with the aforementioned GBP study to obtain RV ejection fraction and LV ejection fraction measurements. For the first-pass acquisition, the gamma camera is placed in the shallow right anterior oblique (20 to 30 degrees) position to maximize the likely separation of the right atrium and RV outflow tract. The radiotracer is injected over several seconds to allow more heartbeats for RV analysis. A large peripheral vein (antecubital or external jugular) is used, and radiotracer administration is followed by a saline flush.

First-pass images for RV evaluation typically are acquired as high temporal resolution list mode data. The bolus is viewed in cine mode to select the appropriate frames for analysis. RV analysis requires isolation of the specific heartbeats (typically three to six beats) during which the bolus first traveled through the right ventricle. For this imaging application, the very low background activity produces high image contrast and reduces the effect of boundary variability on the RV ejection fraction calculation. The selected heartbeat data are binned with respect to the phase of the ECG cycle, and background-corrected RV end-diastolic and end-systolic counts are used to determine the RV ejection fraction. Spatially smoothed data are displayed in cine format for visual and quantitative analysis (Fig. 56-5).