The review and interpretation of myocardial perfusion images are perhaps the key duty of a nuclear cardiologist. It is critical that image interpretation be performed in a systematic fashion so as to maximize the clinical value of the study and to ensure the highest-quality result of the entire procedure, providing optimal clinical information and assisting in clinical decision making. As discussed extensively in Chapter 5, the quality of the study must be reviewed and technical abnormalities be recognized. A comprehensive evaluation of all available imaging data must then be performed so as not to exclude potentially vital information.
A number of guidelines and tools have been recommended for the interpretation of myocardial perfusion studies.1–5 These policies and guidelines have been developed by experts in the field and should be used as a guide to the successful interpretation of myocardial perfusion imaging (MPI). This chapter will provide suggestions for approaches for interpretation based upon these recommendations. Of note, as single-photon emission computed tomography (SPECT) imaging is performed in the vast majority of patients undergoing radionuclide imaging, this chapter focuses on the tomographic evaluation of perfusion and function with SPECT MPI, although the methods recommended in this chapter are largely applicable to PET imaging.
The sequence of imaging should include (1) review of the raw planar images, (2) analysis of the tomographic slices, (3) interpretation of gated SPECT data, and (4) incorporation of clinical data (Table 12-1).
Table 12-1Sequence of SPECT Myocardial Image Interpretation ||Download (.pdf) Table 12-1 Sequence of SPECT Myocardial Image Interpretation
|Review raw planar images |
|Evaluate tomographic slices |
|Analyze gated SPECT data |
|Incorporate clinical data |
It is highly recommended that myocardial perfusion images be reviewed on a computer monitor as opposed to x-ray film or paper. While other media may provide useful information, the resolution of a computer monitor screen and the flexibility in adjusting a variety of parameters, including contrast, thresholds, and colors, makes this the medium that is greatly preferred. The practice of interpreting only "hard copy" images is discouraged, especially in view of the dynamic data, which is available by use of a workstation.
The patient's body habitus should be considered when interpreting images, as this information may support the artifactual nature of apparent perfusion defects. Therefore, data regarding height, weight, and gender should be provided to the interpreting physician. Additional details, such as chest and bra size and the presence of a mastectomy or breast prostheses, may also be useful.
A linear color table is recommended for the interpretation of perfusion images. While linear gray scale is preferred and is recommended by many imaging guidelines, other continuous, linear color tables such as hot body/hot iron revised may also be used effectively (Fig. 12-1). A great variety of other color tables are available. It is critical that the interpreter understands the workings of these various color tables. It is usually recommended to avoid color tables with an abrupt transition between each color, for <10% change in tracer activity. Furthermore, it is critical that when a specific color table is used, the scaling should be linear, not exponential, as this will further enhance the appearance of artifacts potentially leading to false-positive results. Irrespective of the color table selected, the most important aspect of the use of these displays is that the operator be very familiar with the one selected. A bar delineating the color table should also be displayed on screen.
Exercise/rest dual-isotope myocardial perfusion images (thallium-201 for rest and technetium-99m-sestamibi for stress) in a 69-year-old man who presented with atypical chest pain. The myocardial perfusion images were felt to be normal except for the presence of soft tissue attenuation in the inferior and infraseptal regions. These images demonstrate the impact of varying tables. (A) Gray scale (exponential). (B) Hot body (or thermal) (exponential). (C) Warm metal (or CEqual) (linear). (D) Warm metal (exponential). As can be noted, certain tables must be used with a linear scale, not an exponential relationship (C and D).
Care should be taken in the review of images to ensure that the particular tomographic slices are aligned. By convention, the stress study is placed in the top row with the resting study below. It is now well accepted that the display of images should be in a specific format, as more than 10 years ago, the Joint Guidelines from the American Heart Association, the American College of Cardiology, and the Society of Nuclear Medicine stated the manner in which SPECT images should be displayed.6
The top row should present the short-axis views, which are obtained by slicing perpendicular to the long-axis of the lower left ventricle. By convention, the septum is on the left, with the lateral wall on the right. The slices should be displayed from apex to base (left to right). The long-axis should also be presented, demonstrating the data by slicing in a vertical plane (vertical long-axis) and a horizontal plane (horizontal long-axis). The vertical long-axis views should be displayed with the septal slices positioned on the left and progressing to the lateral wall on the right. The horizontal long-axis should be displayed with the inferior slices on the left and moving to the anterior location on the right. All images on the SPECT study should be normalized usually to the brightest pixel in the entire left ventricle within a series of slices (stress or rest). This is called series normalization.4 Flexibility in display software should permit such scaling. Individual frame normalization may optimize image quality but lead to erroneous interpretation.
Although image artifacts may be recognized in several ways, the raw (unprocessed) projection images often provide the most useful information. Therefore, a critical aspect of the interpretation of tomographic data is the review of these rotating planar images.3 All modern camera systems permit such a review often on the same screen as the SPECT slices.
Adequate count data are essential to avoid "splotchy" data, which may easily be confused with perfusion defects. The reasons for low count density include: the dose and type of radiotracer, mode of stress, energy window, collimation, and attenuation. Beyond the visual assessment of rotating planar images or tomographic slices, quantitative assessment of the anterior planar project may be performed, with the requirement for peak pixel count of 100 and 200 for thallium-201 and Tc-99m, respectively.4
While a sinogram (Fig. 12-2) or linogram has frequently been used to demonstrate motion and is composed of a sum of the planar data, this method for quality assessment is not ideal and is not recommended. The preferred approach is a review of a cine loop of both the stress and rest planar images usually simultaneously with the tomographic slices. It is often helpful to place a horizontal line beneath the inferior margin of the left ventricle for both rest and stress to better assess subtle but perhaps significant motion.
Example of a sonogram and a linogram. No patient motion is noted on this example as the borders of the lines are without displacement.
A patient motion-related artifact may be present in up to 15% of all SPECT studies.7 A review of the rotating images provides clear evidence when there is patient motion, in either a superior–inferior manner or laterally. If substantial motion is present (≥2 pixels), repeating the image acquisition is recommended. Motion correction algorithms may also be successfully applied,8,9 but usually only when the patient motion is superior–inferior. Correction of lateral or rotational motion is challenging, but has been successfully incorporated into many software packages.
Multidetector systems may demonstrate abrupt motion on review of the rotating images. This is usually caused by temporal factors associated with a two-detector system and the fact that the last frame of acquisition for detector 1 is substantially later than the first frame obtained from detector 2. Thus, gradual motion throughout the acquisition is therefore accentuated when reviewing the rotating images.
The presence of patient motion may produce artifacts and therefore reduce diagnostic accuracy.10 Not only do these artifacts resemble ischemic heart disease, but also patient motion may create the appearance of multivessel disease. "Upward creep" may also be detected by reviewing the rotating images. This phenomenon occurs when imaging is performed soon after strenuous exercise and results from the repositioning of the cardiac structures as the respiratory excursion decreases following exercise. The "upward creep" artifact may be avoided if imaging is delayed for about 15 minutes after exercise. Prominent patient motion may also be noted on the tomographic slices, producing a characteristic defect such as the "hurricane" sign and "flame" that occurs near the apex, as shown in Figure 12-3.11
Hurricane sign, indicative of patient motion, as described in the original report. (Reproduced with permission from Sorrell V, Figueroa B, Hansen CL. The "hurricane sign": evidence of patient motion artifact on cardiac single-photon emission computed tomographic imaging. J Nucl Cardiol. 1996;3(1):86–88.)
The rotating planar images should be reviewed for the presence of abnormal activity beyond the boundaries of the myocardial structures. Skin or clothing contamination may mask or mimic a true perfusion abnormality but should be identifiable on the rotating images. The presence of intense subdiaphragmatic activity, emanating either from the liver or from the gastrointestinal (GI) tract, may confound image interpretation. Once such activity is present, it may cause a negative lobe artifact, also known as a ramp filter artifact. Intense adjacent activity may cause this reconstruction artifact, for which there is no reliable correction, although iterative reconstruction (as opposed to filtered backprojection) may help.4 This type of abnormality may create an artifactual perfusion abnormality or may mask the presence of a true abnormality (Fig. 12-4). Ideally, when substantial activity is noted especially in the liver or adjacent bowel loop, image acquisition should be repeated to eliminate this type of artifact.
Dual-isotope myocardial perfusion images of a patient felt to be at low risk for coronary artery disease. (A) Prominent activity is noted in a bowel loop immediately adjacent to the infralateral wall on the resting images. This is clearly visible on the resting planar image. These images also demonstrate an apparent reversible inferior wall perfusion defect. (Used with permission from Thomas Holly, MD.) (B) The stress and rest images are again shown on the first two rows in the short-axis, vertical long-axis, and horizontal long-axis images. The third row represents repeat image acquisition of the poststress images following food consumption, defecation, and waiting approximately 2 additional hours. With the subdiaphragmatic/bowel loop activity now removed, there is no perfusion abnormality noted in the inferior wall. (Used with permission from Thomas Holly, MD.)
The interpretation of SPECT images should not be restricted only to the myocardium. A variety of neoplastic lesions may also be detected with commonly used radiopharmaceuticals.3,12 These may reflect either primary or metastatic tumors and include the following types of neoplastic growths: lung, breast, sarcoma, lymphoma, thymoma, parathyroid tumor, thyroid abnormality, and kidney and hepatic tumors. Incidentally discovered clinical thyrotoxicosis can be detected by careful evaluation of the rotating planar images and may aid in the early detection of thyroid disease.13
Finally, the rotating images may reveal contamination by the radiopharmaceutical, occurring on either the skin or clothing, which once again may confound the SPECT image interpretation. In addition, it is usually possible to distinguish between a neoplastic growth and contamination by reviewing the rotating images.
Soft tissue, overlying cardiac structures, may confound image interpretation. Breasts/chest soft tissue, or that related to subdiaphragmatic structures, may reduce the specificity for coronary artery disease (CAD) detection and is present in up to 40% of all studies.
Photopenic areas may be noted from overlapping breast tissue even when the size of the breast is relatively small. It is often possible to appreciate where the reduction of photons may occur on the SPECT slices by reviewing the cine images (Fig. 12-5).
An individual frame of the rotating planar images demonstrating prominent soft tissue attenuation from the breast as depicted by the photopenic area (arrowheads). (Reproduced with permission from Hendel RC, Gibbons RJ, Bateman TM. Use of rotating (cine) planar projection images and the interpretation of a tomographic myocardial perfusion study. J Nucl Cardiol. 1999;6(2):234–240.)
In addition, soft tissue attenuation from the diaphragm may obscure the inferior wall, causing a false impression of an inferior wall abnormality. This occurs most commonly in men. Recognition of the superior-placed diaphragm is helpful in the interpretation of images and the enhanced recognition of a potential artifact. When such an abnormality is present, prone imaging may be helpful. In patients with large BMI, whose defect in the apical-inferior region may be diagnostically challenged, the addition of prone acquisition has been shown to improve diagnostic confidence, and when applied to stress-only MPI can reduce the need for unnecessary rest scans.14,15
Obviously, gated SPECT16,17 and attenuation correction18 methodologies may also be of substantial use in the correct interpretation of soft tissue abnormalities. Differential soft tissue attenuation may occur when the overlying soft tissue is present in different positions on the rest and stress images, thereby leading to the appearance of an apparently reversible perfusion defect. This challenging scenario is not helped by gated SPECT, as true reversible defects (ischemia) may demonstrate normal left ventricular (LV) function.
Analysis of Planar Images
Although tomographic MPI is considered the preferred modality for assessment of myocardial perfusion, planar imaging may be an alternative option in certain circumstances, such as in claustrophobic or critically ill patients where rapid acquisition is required, or in morbidly obese patients that do not qualify for the SPECT camera table. ECG-gated planar images can also be acquired. Regardless of whether the tomographic MPI is performed, planar images should always be inspected first preferably on a linear gray scale. Soft tissue attenuation by breast tissue, diaphragm, or other sources should be noted. Breast marker has been shown to be useful in identifying true perfusion defects from breast attenuation on planar images.
Similarly to the analysis of tomographic slices, segmental analysis of myocardial perfusion can be performed. The standard views for imaging positions and standardized nomenclature for myocardial segmental perfusion evaluation on planar images have been described in the ASNC myocardial perfusion planar imaging guideline.19 For qualitative assessment, the severity of perfusion defect can be classified as mild, moderate, or severe and the extent of defect as small, medium, or large. A five-point segmental scoring system can be applied for semiquantitative evaluation, which is further described later in this chapter. If a quantitative analysis is to be performed on planar imaging, background subtraction must be applied to the images. Reversibility may also be reported from planar images.
Analysis of Tomographic Slices
The first task is to determine whether or not adequate count statistics are present. A number of quality assurance tools are available from most manufacturers that assist in this process. The study should be graded based on overall image quality (uninterpretable, poor, fair, good, and excellent). Factors related to body habitus should be considered.
Cardiac and Lung Activity
The projection data also provide an assessment of cardiac size. Left ventricular hypertrophy (LVH) may be suspected when a reduced LV cavity:wall thickness ratio is noted. Prominent right ventricular uptake may be noted on the raw images or tomographic slices and may indicate right ventricular hypertrophy such as seen in pulmonary hypertension. However, no criterion other than subjective visual impression is available for this diagnosis.
Prominent lung activity may be present, which frequently is present in the setting of severe LV dysfunction or extensive ischemia. However, while abnormal lung activity is an important finding associated with thallium-201 scintigraphy, there is no consensus as to its meaning with technetium-99m imaging.4
LV cavity size may be assessed first by reviewing the rotating planar images. However, the overall cavity-to-wall-thickness ratio may be qualitatively determined by looking at the SPECT slices. In addition, it should be noted if the poststress images reveal a larger LV cavity than noted on the resting study (Fig. 12-6). This would be consistent with transient cavity dilation (TCD) also known as transient ischemic dilation (TID) of the LV cavity. The presence of TCD is a marker of proximal LAD and/or multivessel disease and a worsened prognosis.20 The upper limits of normal TID values vary by protocols and types of isotopes used in the study (Table 12-2). Usually, about a 20% increase is required when using the dual-isotope protocol.20 When using a single-isotope study, a lesser amount of cavity enlargement (approximately 5–10%) is felt to be abnormal.21,22 Pharmacological MPI typically results in higher upper normal TID ratios when compared to exercise MPI as illustrated in Table 12-2. Gender difference for TID thresholds has also been noted, and should be taken into consideration when interpreting TID ratios.23 In addition to the visual assessment, quantitative analysis of the TID ratio is available on most software packages.
Perfusion images from a 31-year-old man with new onset of chest pain, who had limited exercise capacity and developed marked ST-segment changes during the stress test. The stress images reveal an extensive, severe defect in the anterior, septal, and apical regions, with substantial reversibility noted on the resting images. In addition, there is transient enlargement of the left ventricular cavity on the poststress images, relative to the resting study. The transient cavity dilation is most notable on the vertical and horizontal long-axis images; the TID ratio was 1.4.
Table 12-2Abnormal TID Cutoffs for Different Protocols and Types of Isotope Used ||Download (.pdf) Table 12-2 Abnormal TID Cutoffs for Different Protocols and Types of Isotope Used
|Protocol ||TID Threshold |
|Rest Tl-201/exercise stress Tc-99m sestamibi20 ||1.22 |
|Rest Tl-201/exercise stress Tc-99m sestamibi24 ||1.23 |
|Rest Tl-201/pharmacologic stress Tc-99m sestamibi || |
| Dipyridamole25 ||1.27 |
| Adenodine25 ||1.35 |
| Adenosine26 ||1.36 |
| Regadenoson27 ||1.39 |
| Dobutamine25 ||1.40 |
|Exercise stress/rest Tc-99m sestamibi24 ||1.14 |
|Rest/exercise stress Tc-99m sestamibi21 ||1.19 |
|Gated rest/exercise stress Tc-99m sestamibi (end-diastolic volume)21 ||1.23 |
|Regadenoson stress/rest Tc-99m sestamibi28 ||1.33 |
|Rest/regadenoson stress Tc-99m tetrofosmin22 ||1.31 |
|Dipyridamole stress/rest Tc-99m sestamibi (2-day)29 ||1.19 |
|Gated stress/rest Tc-99m tetrofosmin (2-day; end-diastolic volume)30 ||1.25 |
Defect severity is often described in a qualitative fashion (mild, moderate, and severe). A mild abnormality is one in which the clinical significance of the defect is unknown. Such an abnormality may reflect an equivocal finding. This often represents only a 10% reduction of peak tracer activity for a particular study. Moderate and severe defects carry more important diagnostic and prognostic value. In addition, the extent of the perfusion abnormality may also be qualitatively described as small, medium, or large (Figs. 12-7 and 12-8). Although these descriptions are relative, they may be based on objective information from quantitative programs.
Exercise/rest dual-isotope myocardial perfusion images from a 69-year-old man with a history of hypertension, hyperlipidemia, and diabetes who presents with exertional chest pain. There is a small-sized defect of moderate severity involving the basal portion of the inferior wall. This perfusion abnormality appears completely reversible and is consistent with the significant stenosis in the right coronary artery. Subsequent coronary angiography confirms the presence of a 95% right coronary artery stenosis.
Adenosine/rest dual-isotope myocardial perfusion imaging in a 75-year-old man with a history of known coronary artery disease and status postmyocardial infarction. He is currently asymptomatic. The perfusion images demonstrate a large area of severely reduced activity in the inferior and inferoseptal walls, with a moderately severe abnormality noted in the septum and apical regions. No reversibility (ischemia) is noted.
In an attempt to describe the severity and extent as a combined value, a variety of scoring systems have been designed, the most popular being the summed stress and summed rest scores. These scores are derived by adding the point value using the range of "0" for normal perfusion to "4" for absent activity for each segment of the 17-segment model.1 A mild, moderate, or severe reduction in count should be scored as 1, 2, or 3, respectively. The difference between the summed stress score and the summed rest score is called the summed difference score and is a measure of reversibility. Usually, individual segments with a ≥2-grade improvement on the resting study are felt to represent substantial ischemia. The size of the defect should be noted, as small, moderate, or large (Table 12-3).
Table 12-3Semiquantitative Defect Analysis ||Download (.pdf) Table 12-3 Semiquantitative Defect Analysis
| ||Percent of Left Ventricle ||Number of Segmentsa |
|Small ||5–10 ||1–2 |
|Moderate ||15–20 ||3–4 |
|Large ||>20 ||≥5 |
The type of perfusion abnormality should also be described. A fixed perfusion defect (i.e., one that is the same on both the post stress and rest images) is often equated to a myocardial scar, especially when the abnormality is of severe intensity. However, a fixed perfusion abnormality may also reflect severe myocardial ischemia and the presence of myocardial viability. A reversible abnormality is a perfusion abnormality noted on the poststress images, but largely normalizes on the resting images. In many cases, some interpreters may use the term partially reversible. It is critical to determine whether it is a predominantly reversible defect or only a minimally reversible abnormality. Quantitatively, reversibility has a variety of definitions, but is often associated with a 20% to 30% improvement in regional activity.
"Reverse redistribution" describes a pattern where a defect noted on the rest images is either not present or less severe on the stress images. This finding may be noted in the setting of a subendocardial (nontransmural) scar and provides evidence of viability.31 Although well described with thallium-201 scintigraphy, when this pattern is present with Tc-99m sestamibi or tetrofosmin, an artifact should be suspected. This finding likely relates to low count density of the resting study especially with a same day rest/stress protocol and does not correlate with significant coronary artery lesions.32
The perfusion abnormalities should also be identified by their location. Standard terminology has now been accepted.1 The 17-segment model should be used for reference with regard to the nomenclature of such abnormalities (Fig. 12-9). However, in general terms, perfusion abnormalities should be described as being present in the apical, anterior, inferior, or lateral walls/regions. The term "posterior" should not be used. The perfusion abnormality may also be described as occurring within a specific vascular distribution. Obviously, the distribution of an individual coronary artery is highly variable. However, by convention, the 17 segments have been assigned specific vascular distributions so as to standardize interpretation and reporting. As a general rule, the lateral wall is assigned to the circumflex distribution, the anterior and anteroseptal regions to the left anterior descending coronary artery, the inferoseptal and inferior walls to the right coronary artery, and the apex is usually assigned to the left anterior descending distribution, although this is highly variable.
(A) A polar-plot depiction of left ventricular segmentation according to the 17-segment model. The recommended nomenclature is noted for each segment below. (B) The 17-segment model, obtained by three individual short-axis slices as well as one mid-cavity vertical long-axis slice. A depiction of the coronary artery distribution is also noted. (Reproduced with permission from Cerqueira MD, Weissman NJ, Dilsizian V, et al. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: a statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. Circulation. 2002;105(4):539–542.)
A variety of software tools are presently available, including the following products that are commercially available. These quantitative programs usually reference an individual patient's data to a normal reference profile. The comparison of individual studies to such a reference is often displayed as a polar map. The "blacked-out" segments usually reflect an area of activity that is below the threshold deemed as normal (Fig. 12-10). In many cases, it may represent a value such as 2.5 standard deviations below the mean value for a normal patient population; individual programs have specific thresholds. These thresholds and normal reference files are often different depending on the radiopharmaceutical. Furthermore, additional techniques, such as attenuation correction, may also alter the profiles. Most important, however, is that the normal reference files are gender specific unless attenuation correction methodology is employed. In addition to the polar map or bull's-eye projection, circumferential profiles may also be created again demonstrating where the count density falls below a specific threshold and is, therefore, deemed abnormal.
Stress/rest perfusion imaging demonstrates a large inferior, interoseptal, and interolateral defect, which has a small area of reversibility (ischemic) in the apex. The polar plots demonstrate the large extent of the defect, with the hatched region indicating reversibility.
Quantitative analysis for most of the software programs has been validated in multiple studies and usually published in peer-reviewed journals,33,34 which is further discussed in Chapter 9. However, it is advised that the quantitative analysis be used as a tool and guide, serving as a "second observer." These quantitative computer-assisted tools should not be used for primary analysis. Following the visual inspection of the tomographic slices, quantitative interpretation may be examined. Any discrepancies or previously unrecognized abnormalities may then be reviewed. However, the individual interpreter must "overread" the computer-assisted interpretations, as many technical problems may develop and lead to false results. Therefore, quantitative analysis is not a substitute for an expert interpretation but should be used as an adjunct to assist the interpreter.
It is now recommended that gated SPECT be employed for essentially all myocardial perfusion SPECT imaging studies35 with a standardized approach for the interpretation of gated SPECT data. This critical component of contemporary perfusion imaging is discussed in detail in Chapter 11. The gated SPECT data are often displayed in different fashions, depending on the software. Irrespective of the display, the most critical information is often demonstrated in the mid-ventricular slices from each of the orthogonal axes. Gated SPECT should be displayed as a cine loop, and the interpreter should observe the images for overall global function, examining the endocardial surfaces and their excursion. In addition to the motion of the endocardial walls, myocardial thickening may be determined by the increase in brightness noted on gated SPECT display resulting from the partial volume effect. It is possible that the thickening (brightening) may be normal, although the excursion is abnormal. This is often seen in settings such as following previous cardiac surgery, where the septum appears dyskinetic or akinetic, but thickens (brightens) normally. If there is difficulty localizing the endocardial surfaces, most software will provide "contours" where the computer will provide a line for what it believes to be the endocardial surface (Fig. 12-11). This may be used to assist the interpreter in evaluation of endocardial motion. Regional abnormalities may also be determined especially by examining multiple axes. Similar geographic schema to that noted for perfusion imaging should be employed when describing regional wall motion abnormalities.
Computer-assisted detection of the endocardial surfaces on gated SPECT with the "contours" demonstrating the boundaries of the endocardial and epicardial surfaces (4 D-M SPECT™).
A great variety exists with regard to the use of displays for gated SPECT information. While black and white or "hot body/thermal" may demonstrate brightening very effectively, a number of different color tables have also been used to assist in evaluating myocardial brightening or increases in count intensity. Overall, however, the monochromatic color tables are usually recommended.
It is critical to have sufficient count density to examine for the accurate interpretation of gated SPECT data. A number of software programs analyze each of the frames to determine if the count density is adequate. The overall image quality should help the interpreter determine whether or not the study is interpretable. Another concern is that of poor gating, which is manifested as a flashing on the rotating planar images. This is due to variation in the beat–beat interval. For the most part, however, the overall global function data, including the ejection fraction, is still well preserved. If anything, cardiac arrhythmias more often affect the myocardial perfusion data than they do the gated SPECT information and its impact on functional information. The use of a heart rate histogram may be useful in this setting to explain apparent count "drop out" (Fig. 12-12).
Histogram of patients with a regular rhythm and narrow range of beat rate (Panel A) and with an irregular rhythm with a substantial portion of beats lying outside of the R–R interval window (Panel B).
Global and regional wall motion abnormalities should be defined as normal, hypokinetic, akinetic, or dyskinetic. It is possible to further subdivide the hypokinesis, although it may be difficult to differentiate between mild and moderate hypokinesis. A five-point scoring system for thickening and wall motion has been described ranging from normal function to mild or moderate hypokinesis, to akinesis and dyskinesis. Usually, wall thickening correlates well with wall motion. However, following cardiac surgery, there is usually reduced excursion of the septum, with normal thickening, a finding that is a normal variant.
Global LV function may be accurately quantified and described specifically using an ejection fraction determination. Each software program has been well validated and reveals good correlation with other methodologies. When the ejection fraction is >70%, such as occurring in patients with small LV cavities, it is suggested to describe this as either "normal" or "≥70%," as it is somewhat nonsensical to describe an ejection fraction of 92%. More qualitative descriptions may also be used such as "normal function" or those studies possessing mild, moderate, or severely reduced LV systolic function. However, given the overall validation of cardiac software packages, it is recommended to quantitatively describe the ejection fraction. LV volumes, both end systolic and end diastolic, may also be noted.
Regional wall motion abnormalities and regional myocardial thickening have also been accurately determined using several cardiac software packages. This information is often graphically depicted as a three-dimensional plot. This may be used to assist the interpreter in determinations of abnormal function. However, most of these methods have been less well validated than the global ejection fraction determination. Therefore, the tools for regional determinations should be used as an adjunctive technique to assist the interpreter.
A number of manufacturers possess well-validated methods for correcting soft tissue attenuation.18 Although different techniques have been employed by most vendors, the literature now supports the conclusion that diagnostic specificity is improved. In addition, it appears that attenuation correction may assist in the improved detection of multivessel disease and left main stenosis. It is also likely that attenuation correction will assist in prognostic applications. It is, however, critical to understand the workings of each system, as they are widely different. The interpreter of myocardial perfusion SPECT imaging should note each system's benefits and potential limitations. All attenuation correction methods may cause artifacts, especially when used incorrectly. Therefore, the interpreter should know the specifics of such attenuation correction–derived artifacts as well as how effective it is in correcting soft tissue attenuation.
It is critical that the quality of the transmission map used for correcting the emission data be of high quality (Fig. 12-13). The counts should be adequate, as determined by the manufacturer, and truncation should be absent or minimal. If the quality of the transmission scan is suboptimal, the attenuation-corrected images should not be used.
Stress images obtained in a 49-year-old man with a low likelihood for coronary artery disease. The top rows demonstrate an inferior and interoseptal wall perfusion abnormality. Following attenuation correction (Vantage Pro™, Philips), a more uniform appearance is present (bottom rows). The bottom portion of the figure depicts the transmission map, which clearly demonstrates the lungs and mediastinal structures. (Used with permission from Gary Heller, MD, PhD.)
It is presently advised that the attenuation-corrected images be viewed in conjunction with the review of the uncorrected images. It is critical to understand how the correction occurred and its impact on the images. With the understandings of the benefits and limitations of each system, as well as comparing the uncorrected and corrected information, the interpreter can then gain the true value of attenuation-corrected perfusion imaging. Caution is often advised when there is prominent activity from overlying structures such as the bowel or liver. This can directly impact on the interpretation of an inferior wall abnormality. Specifically, if substantial hepatic or subdiaphragmatic activity is present, a true inferior wall perfusion abnormality may be masked by attenuation correction. Importantly, apical thinning is also far more prominent on attenuation-corrected SPECT images (Fig. 12-14). It is, therefore, critical to understand the "normal" appearance of an attenuation-corrected image. Likewise, the right ventricle is far more prominent on attenuation-corrected SPECT imaging. This does not reflect right ventricular enlargement or hypertrophy, but instead improves visualization of the structure. Obviously, a "learning curve" is required to understand the impact of attenuation correction on the LV apex and right ventricle. It is recommended that the uncorrected SPECT images be examined in addition to the attenuation-corrected data.4,5
Horizontal long-axis images of a patient with normal myocardial perfusion. The top two rows depict standard, noncorrected stress and rest SPECT images, with the pink arrow pointed at the apex. The bottom two rows are following attenuation correction, demonstrating an apical defect (yellow arrows), which is due to apical thinning and commonly seen following attenuation correction.
In addition to the commercially available attenuation correction software, combined supine-prone myocardial perfusion SPECT without attenuation correction has been shown to minimize soft tissue attenuation and increase specificity and diagnostic accuracy in the detection of CAD, especially in women and obese patients.14,15 By shifting the heart anteriorly and the diaphragm and subdiaphragmatic organs inferiorly, prone position improves inferior wall attenuation artifact. However, prone-only MPI is not recommended due to possible artifactual anterior and anteroseptal defects from the close proximity of the myocardium to anterior bony structures, for example, ribs and sternum, resulting in false-positive results.
The use of combined supine-prone MPI has similar prognostic values compared to attenuation-corrected MPI. The presence of perfusion defects on supine acquisitions but normal on prone images indicates a good prognosis and a low risk for subsequent cardiovascular events, similar to that of patients with normal supine-only studies.14,15
Incorporation of Clinical Data
Beyond information regarding the patient's body habitus, clinical data, including risk factors, may influence image interpretation. As such, these data should be considered only after careful image interpretation. How this information should be weighed with regard to final impression is controversial, although most experts recommend not including this in the final conclusion. However, data regarding prior cardiac events, including revascularization and myocardial infarction (MI), must be considered so as to add to the clinical relevance of the study.
Clinical Decision Support Systems
A clinical decision support systems (CDSS) is an interactive computer software developed to assist physicians and health care professionals with interpretation of imaging studies and to support clinical decision making in order to enable faster interpretation and more accurate diagnosis. With the aid of an artificial intelligence, subjectivity and intra- and interobserver variation in image interpretation are minimized, resulting in standardized high-level performance and more cost-effective care.
A CDSS has two basic components: (1) a dynamic knowledge base obtained from evidence-based medicine and expert opinions, and (2) an inferencing mechanism which is a navigational tool to reach a conclusion. Various methodologies and techniques used in MPI CDSS have been well described, such as bayesian networks, artificial neural networks, case-based reasoning, rule-based systems, data mining, and natural language processing.36 After the images and clinical data are processed, a structured report is generated by the system, which allows the interpreter to edit or override the automated results from CDSS. The report generated by CDSS should meet the standardized reporting guidelines as further described in this chapter.
Despite the application of artificial intelligence to nuclear imaging, human interpreters will remain the role of primary diagnosticians, and results should be reviewed and confirmed by physicians. The development of integrated CDSS software will continue to ensure that nuclear cardiology remains the leader in the field of digital imaging.
A systematic process of interpretation of myocardial perfusion SPECT images is critical for the highest level of accuracy and clinical utility. Attention must be paid to the quality of MPI data. In recent years, a variety of guidelines and position papers have focused on optimal techniques for image interpretation.1–5,7,14,15,17–19 A complete listing of all items to be considered, including those that are standard, recommended, or optional, may be found in the most recent version of the ASNC guidelines for image interpretation.4,5,19 A common vocabulary and format also ensures that high-quality interpretation and reporting is achieved. Advanced imaging technology, including quantitative analysis, gated SPECT, and attenuation correction, has improved the value of SPECT imaging, but the "reader" of such images must be cognizant of the advantages and pitfalls for these methods and understand the potential impact of these techniques on the final interpretation.