The noninvasive assessment of resting left ventricular (LV) performance is an integral part of the evaluation of patients with known or suspected cardiac disease, having important diagnostic, therapeutic, and prognostic significance.^1–11 Although scintigraphic measures of cardiac function historically included measurement of ejection fraction (EF), estimation of cardiac output, valvular regurgitant fraction, and detection of intracardiac shunts have also been successfully performed. Other than EF, these potential applications for nuclear cardiology techniques have been largely supplanted by echocardiography, cardiac computed tomography, and magnetic resonance imaging (MRI) techniques over the past two decades.

Gated equilibrium radionuclide angiography (ERNA, often called multiple-gated acquisition [MUGA] and equilibrium radionuclide ventriculography [RNV]) were introduced nearly 30 years ago. These techniques were routinely utilized in the evaluation of patients with known or suspected LV dysfunction, postmyocardial infarction (MI), and valvular disease, and for monitoring the cardiotoxic effects of chemotherapeutic drugs. Exercise radionuclide angiography (RNA), particularly with the first-pass radionuclide angiographic (FPRNA) technique, was widely used to diagnose or evaluate known coronary artery disease (CAD),^5–9 competing favorably with and often complementing planar myocardial perfusion imaging (MPI). However, since its introduction, RNA has evolved little, while other noninvasive methods, such as gated single-photon emission computed tomography (SPECT) MPI, echocardiography, and cardiac magnetic resonance, have been introduced and become increasingly sophisticated and cost effective.

Red Blood Cell Labeling

For FPRNA alone, Tc-99m DTPA is often used if no equilibrium images are required. For ERNA, red blood cell labeling can be performed via multiple methods. The in vitro method, usually using a commercial kit, is preferred, and offers the highest level of labeling efficiency (>97%). Blood is withdrawn from the patient, mixed into a “reaction vial” containing stannous pyrophosphate. After 5 minutes it is mixed with sodium hypochlorite to destroy extracellular stannous ion, and then with citrate buffer. After shaking lightly, 25mCi Tc-99m is added, incubated for 20 minutes and then reinjected back into the patient. The in vivo method, although simple, cheap and fast, is no longer recommended as it provides the lowest labeling efficiency (80–85%). It involves giving stannous pyrophosphate directly to the patient, followed by Tc-99m 15 to 20 minutes later (which can be given as a rapid bolus if for FPRNA). Finally, a modified in vivo/in vitro method can be used with labeling efficiency of 92% to 95%. Perfusion imaging can also be performed if desired.^12–14

First-Pass Radionuclide Angiography

FPRNA images are usually obtained using a single- or multicrystal high-count rate gamma camera fitted with a high-sensitivity parallel hole collimator (e.g., SIM 400, Scinticor, Milwaukee, WI; or ElGems CardiaL [formerly Elscint], Haifa, Israel). Images are acquired in the anterior or the right anterior oblique (RAO) projection using 25 (±4) frames per cardiac cycle. Preliminary work suggests that FPRNA can also be performed with positron emission tomography (PET) using images acquired during a bolus of radiolabeled water (H₂¹⁵O), but this has not yet achieved widespread utilization or validation.¹⁵

FPRNA data are analyzed using the frame method for LVEF using commercially available computer software,^16–18 as shown in Figures 22-1, 22-2, 22-3. This software creates a representative LV volume curve by summing frames of several (usually 5–10) cardiac cycles, which are aligned by matching their end diastoles (histogram peaks) and end systoles (histogram valleys) during the operator-defined levophase of radioactive tracer transit. The pulmonary-frame background-corrected representative cycle is then examined with a fixed region of interest (ROI) in order to obtain the final first-pass LV time–activity curve. This ROI is drawn over the LV as defined by a first harmonic Fourier transformation phase image, which distinguishes clearly the LV from the aortic counts. End diastole is taken as the first frame of the representative cycle, and end systole is defined as the frame with the minimum counts in the histogram. Historically, the LVEF was taken as the end-diastolic counts minus the end-systolic counts, divided by the background-subtracted end-diastolic counts.

A second ROI (end diastolic) can be derived from a Fourier transformation amplitude image with masking of the lower 10% of image intensity, which extends the ROI in a basal direction, usually one to four pixels, depending on the vigor of ventricular contraction, up to the amplitude signal of the aortic root. The remainder of this ROI is drawn to match the first ROI (end systolic). The dual ROI LVEF is determined as the end-diastolic ROI counts minus the end-systolic ROI counts, divided by the background-subtracted end-diastolic ROI counts. This results in an accounting for valve plane motion during the cardiac cycle, using Fourier-guided dual ROI analysis of FPRNA, giving EFs that are highly reproducible and similar in value to ERNA.¹⁸

Gated Planar Equilibrium Blood Pool Imaging

Electrocardiogram (ECG)-gated planar equilibrium blood pool images (Fig. 22-4) can be performed with any of the several blood pool agents, such as Tc-99m–labeled red blood cells (described above and the standard) and Tc-99m–labeled human serum albumin. These images are best when acquired using high-resolution collimation. Images for planar LVEF calculation are obtained in the best septal (shallow) left anterior oblique (LAO) view. For regional wall motion assessment, the best septal view ±45 degrees should be obtained (approximately “anterior” and “lateral” views). Each of these planar-gated images should be acquired for 6- to 10-minute duration, dividing each cardiac cycle into a minimum of 32 frames. A smaller number of frames (e.g., 16) are adequate for assessment of EF due to the usual length of the isovolumic ejection period (at end diastole) and relaxation period (at end systole). However, more frames are desirable if quantitative assessment of diastolic performance is needed.

For LVEF, automated regions of interest are generated on the planar-gated equilibrium blood pool data throughout the cardiac cycle using the commercially available software. An automated periventricular background ROI must be adjusted routinely in order to avoid inclusion of high-count structures (e.g., spleen and descending aorta), which would artifactually increase the LVEF if included.

Gated Tomographic Equilibrium Blood Pool Imaging

ECG-gated projections for SPECT reconstruction are usually obtained using high-resolution collimation in a fashion similar to gated SPECT perfusion imaging (Fig. 22-5), scanning from RAO 45 degrees to left posterior oblique (LPO) 45 degrees projections. A total of 45 to 60 projections of 20- to 30-second duration, at 4-degree or 3-degree steps, respectively, should give adequate count density. At each projection, either 8 or 16 frames per cardiac cycle should be acquired. After collimator sensitivity and center of rotation correction, projections are reconstructed into transaxial slices of the blood pool for each of the 8 or 16 frames of the cardiac cycle. The transaxial slice sets can then be reoriented in cardiac planes (i.e., short axis, horizontal long axis, and vertical long axis) for each of the eight frames of the cardiac cycle.

Multiple methods of quantifying blood pool SPECT images have been published using either three-dimensional images or midventricular horizontal and vertical long-axis slices, which can be analyzed for wall motion or EF.^19–25 The LVEF and RVEF can be calculated for each long axis as the end-diastolic counts minus the end-systolic counts, divided by the end-diastolic counts, combining them for a biplane EF.

The three-dimensional techniques are useful for blood pool volume rendering, which is used for delineation of myocardial (actually endocardial) topography. This technique is limited by the accuracy of automated valve plane definition since there is often little count density change at the mitral or tricuspid valve planes. However, this method has been validated against standard approaches, such as cardiac magnetic resonance.²⁵ Due to the uniformly high contrast and lack of overlap between cardiac blood pool and extracardiac structures, no background subtraction algorithm should be employed for these analyses.

Gated SPECT Myocardial Perfusion Imaging

Gated SPECT MPI (Fig. 22-6) has become one of the most powerful tools available in nuclear cardiology and is discussed in detail in Chapter 11. Acquisition of SPECT data, synchronized with the electrocardiographic R wave, is generally performed using 8 or 16 gating intervals and allows evaluation of both global (EF) and regional (myocardial wall motion and wall thickening) cardiac function. Gated SPECT acquisition is now recommended for all perfusion studies whenever possible. It is estimated that over 90% of all SPECT studies in the United States are currently performed using the gated acquisition technique.²⁶ Gated SPECT MPI is most often performed with poststress ECG gating. Ungated SPECT perfusion imaging is usually reserved for resting images in some institutions or for patients with cardiac arrhythmias.

Two- and three-dimensional techniques for display of gated perfusion SPECT have been well validated for EF.^26–31 Gated SPECT has been utilized extensively for determination of EF and wall motion, adding incremental diagnostic and prognostic information. The addition of wall motion has improved the specificity of myocardial perfusion SPECT by distinguishing myocardial scarring from attenuation artifacts, both of which may result in “fixed” defects.³²

The exponential increase in the use of gated SPECT perfusion has been fueled by the increased availability of automatic and semiautomatic algorithms for the quantification of cardiac function. DePuey et al.²⁹ described a method based on the automated detection of endocardial borders. Williams and Taillon²⁸ developed a method based on digitally subtracting and inverting the perfusion images and analyzing the change in counts occurring in the ventricular chamber, using an edge-detection method identical to that commonly used in gated ERNA. Germano et al.^27,30 developed a totally automated method of fitting geometric shapes to the endocardial borders to obtain systolic and diastolic volumes and EFs. Smith et al.³¹ have used partial volume effect to quantitate regional thickening and estimate LVEF, without need for any edge-detection algorithm. Each of these techniques has been well “correlated” with standard methods of performing EF, but each demonstrating varying degrees of “substitutability.”²⁸

For these reasons, the gated SPECT policy statement of the American Society of Nuclear Cardiology encourages the use of gated SPECT with every perfusion study in which gating is feasible. This reflects the improved quality control and diagnostic accuracy of interpreting perfusion images in light of regional and global LV performance, as well as the incremental value of gated SPECT over perfusion imaging alone for prognosis.^32–35