Gating of Myocardial Perfusion Imaging for the Identification of Artifacts: Is It Useful for Experienced Physicians?

Abstract

Gating of myocardial perfusion imaging helps to differentiate artifacts from perfusion defects. We used this technique to evaluate the impact of routine gating on the interpretation of results by physicians experienced in this field.

We studied, prospectively, 270 consecutive patients (161 men and 109 women) who underwent gated myocardial perfusion imaging. Single-photon emission-computed tomography was performed to evaluate myocardial perfusion in patients at rest and after stress, using technetium-99m sestamibi and post-stress gating. Participating physicians interpreted each study and indicated a confidence level for the interpretation. Initially, these opinions were formed on the basis of static slices alone and subsequently, with the addition of gating information. The impact of gating was evaluated by the number of studies in which gating led to a change in interpretation from normal to abnormal or vice versa, or from borderline to definite.

The interpretation was changed from abnormal to normal or vice versa in 10 studies (3.7%) and from borderline to definite in 3 (1.1%). In 37 studies (13.7%), the confidence level was increased from confident to very confident with no change in interpretation.

We conclude that routine gating of every myocardial perfusion imaging study for the identification of artifacts is of low value for physicians experienced in interpreting such studies. Although gating frequently increases the confidence level, it seldom leads to a change in interpretation. Specific subgroups of patients who would benefit from gating should be identified.

Myocardial perfusion imaging (MPI) is widely used in the diagnosis of myocardial perfusion. ^1–4 Soft-tissue attenuation has long been recognized as 1 of the most frequent causes of artifacts in MPI. ^5–7 Gating has been used as a tool for differentiating between true perfusion defects and artifacts, and it has been shown to decrease the number of studies interpreted as borderline and to improve the accuracy of the study. ^8,9 However, the number of studies that are interpreted as borderline depends very much on the experience and confidence levels of the interpreting physician. In this study, we used routine gating with every MPI study in order to evaluate the impact of gating on the interpretation and confidence levels of physicians who had extensive experience in reading such studies.

Patients and Methods

Two hundred seventy consecutive patients who underwent gated myocardial perfusion imaging in our department (161 men and 109 women) were prospectively studied. Gating was not performed in patients with severe arrhythmia; therefore, those patients were excluded from the study.

In most patients, myocardial perfusion imaging was performed with technetium-99m (Tc-99m) sestamibi using a single-day rest-stress protocol; ¹⁰ in severely overweight patients, a 2-day protocol was performed. For rest imaging, 370 megabequerels (MBq) of Tc-99m sestamibi was injected intravenously for the 1-day protocol, and a 999- to 1110-MBq injection was used for the 2-day protocol. One hour after injection, single-photon emission-computed tomography (SPECT) was performed with either: 1) a triple-detector camera (Prism 3000 XP; Picker, Cleveland, Ohio), a low-energy, high-resolution collimator, a 20% symmetrical window at 140 kiloelectron volts (keV), a 64 × 64 matrix, an elliptical orbit with 120 projections, with step-and-shoot acquisition at 3-degree intervals and a 20-sec dwell time per stop; or 2) a single-detector camera (General Electric Starcam; Milwaukee, Wis), a low-energy, high-resolution collimator, a 20% symmetrical window at 140 keV, a 64 × 64 matrix, a circular orbit with 64 projections, with step-and-shoot acquisition at 3-degree intervals and a 20-sec dwell time per stop.

After the completion of rest imaging in the 1-day protocol, or on a separate day in the 2-day protocol, stress was induced by 1 of these methods: treadmill exercise according to the Bruce protocol; dipyridamole (0.57 mg/kg injected intravenously over 4 minutes); or dobutamine (a starting dose of 10 mcg/kg/min with incremental increases to a maximum dose of 50 mcg/kg/min). For stress imaging, 999 to 1110 MBq of Tc-99m sestamibi was injected intravenously. Imaging was performed 15 to 45 minutes after the end of stress according to the SPECT imaging protocol described above. All stress acquisitions were gated at 8 frames per R-R cycle with a 20% window.

Image Interpretation

All studies were interpreted by 1 of 3 nuclear medicine specialists, each of whom had at least 7 years of experience in interpreting MPI studies in a nuclear cardiology laboratory. These physicians had used gating routinely in the interpretation of MPI during the previous 5 years. For the purposes of this study, the physicians were blinded to the patients' clinical information except for sex, breast size, weight, and height.

The static tomographic perfusion images were viewed first, followed by the gated images. The static images were displayed in the short-axis, vertical long-axis, and horizontal long-axis views. The interpreter was allowed to modify the alignment of the images, the window, and the intensity, in order to select the most appropriate display of static slices. The short-axis and vertical long-axis myocardial tomograms were divided into 15 segments for each study (Fig. 1). In turn, these segments were assigned to 4 regions in the apical short-axis slice, 5 regions each in the midventricular and basal short-axis slices, and 1 apical region in the midventricular, vertical long-axis slices.

Fig. 1 Assignment of myocardial segments in the single-photon emission-computed tomography (SPECT) images.

A = apical short-axis view; B = mid-ventricular short-axis view; C = basal short-axis view; D = mid-ventricular vertical long-axis view

Reviewing the static tomographic slices, the interpreters rated the perfusion of the segments for both stress and rest using a 5-point scoring system (0 = normal, 1 = mildly decreased, 2 = moderately decreased, 3 = markedly decreased, and 4 = absent). Defects with similar perfusion at stress and rest were designated as “fixed”; defects at stress with complete normalization at rest, “reversible”; and defects at stress with partial improvement at rest, “mixed.” After rating the perfusion of each segment, the physician assigned a confidence level to his or her interpretation of each segment, from A to D (A = very confident, B = confident, C = suspicious, D = not confident). These levels described the certainty of the interpreter regarding the perfusion of each segment. At the end of the evaluation of the static slices, the entire study was characterized as having fixed, reversible, or mixed defects. A confidence level from A to D (as defined above) was then assigned to that overall impression.

Subsequently, the gated slices were displayed in a cine mode in the short-axis, horizontal long-axis, and vertical long-axis views. The interpreter was able to choose from all the gated slices in these 3 axes. As with the static slices, the gated images were divided into 15 segments. For each segment, the wall motion and thickening were visually evaluated. If either was abnormal, the segment was considered abnormal by gated analysis; if both were normal, the gating of the segment was considered normal. The interpreter then assigned a confidence level (A to D) to each segment according to his or her impression of the gated analysis. At the end of the gating evaluation, the wall motion and thickening of the left ventricle were described as a whole as normal or abnormal, and a confidence level from A to D was assigned to that interpretation as well. Finally, a combined impression from both the static and the gated information was established for every segment and then for the entire study. Therefore, each segment was designated as normal, fixed, reversible, or mixed, with a certain confidence level; and each study as a whole was interpreted as involving fixed, reversible, or mixed defects—or a combination of those—with a certain confidence level for this designation. Any interpretation with a confidence level of A or B was considered “definite” (normal or abnormal), and any interpretation with a confidence level of C or D was considered “borderline.” After the evaluation of the gated slices, the unprocessed data were reviewed in a cinematic format to identify any motion during the imaging or any soft-tissue attenuation (mainly from the breast).

Gating was considered to have influenced the interpretation of myocardial perfusion images if 1 of the following occurred:

• the initial interpretation of a study or a segment (based only on the static slices)—when compared with the final interpretation (based on both the static and the gated slices)—was changed from abnormal to normal or vice versa

• an initial confidence level in the interpretation of a study or a segment was changed from borderline (C or D) to definite (A or B)

• the confidence level of the initial interpretation of a study or a segment, when compared with the final interpretation, was raised from not confident to suspicious (D to C) or from confident to very confident (B to A) without a change in the interpretation.

Statistical Analysis

Intergroup comparisons were performed with the χ² test for categorical variables. Continuous variables were expressed as mean ± standard deviation. P values <0.05 were considered significant.

Results

Table I summarizes the demographic, stress, and myocardial perfusion characteristics of our study group. Of the 270 studies, the interpretation was changed due to gating information from abnormal to normal or vice versa in 10 (3.7%). In 7 of those (2.6%), the interpretation was changed from abnormal to normal, and in the other 3 (1.1%), from normal to abnormal.

Table I. Demographic, Stress, and Myocardial Perfusion Imaging Characteristics

When confidence regarding the 270 studies was examined, results showed that there was an increase in confidence in 40 studies (14.8%). In 37 of these (13.7%), the level was raised from confident (B) to very confident (A). The confidence level was increased from borderline (C) to definite (A or B) in only 3 (1.1%). (No studies were assigned a confidence level of D.) In no case was the confidence level lowered because of the addition of gated information.

Overall, 4,050 segments (270 studies × 15 segments) were evaluated. In only 37 (0.9%) of the individual segments was there a change in the interpretation from abnormal to normal or vice versa. There was an increase in the confidence level in 140 (3.5%) of the 4,050 segments. However, in only 29 of those (0.7%) was there an increase in confidence level from borderline (C) to definite (A or B). In the remaining 111 changed segments, the confidence level was raised from B to A without a change in the interpretation.

Separate Results of Studies in Men and Women

Separate analyses were performed on the results of men's and women's studies with regard to changes in interpretation. Of the 161 studies performed in men, 9 (5.6%) were changed from normal to abnormal or vice versa due to gating information (2, normal to abnormal; and 7, abnormal to normal). In contrast, just 1 of the 109 studies (0.9%) performed in women was altered, and that was from abnormal to normal (P = 0.04) (Fig. 2).

Fig. 2 The percentage of men (black bars) and women (shaded bars) in which the interpretation was changed from normal to abnormal or vice versa (group 1) or in which there was some increase in confidence level (group 2).

When the confidence levels regarding the studies in men were analyzed, results showed that changes were made in 33 (20.5%). Two of those were changed from borderline (C or D) to a stronger confidence level (A or B), with a subsequent change in the interpretation. In the other 31 studies, the confidence level was increased from confident (B) to very confident (A) but without a change in the interpretation.

When the confidence levels for the studies in women were examined, results indicated that levels changed in 7 (6.4%). Of those, there was only 1 in which the confidence level was increased from borderline to definite with a change in the interpretation. In the other 6 studies, the confidence level was increased, but the interpretation was not altered (P <0.001) (Fig. 2).

Discussion

Myocardial perfusion imaging is widely used in the diagnosis of coronary artery disease, as well as for the prediction of patient outcomes. 1–5,11–13 However, several factors affect its accuracy, resulting in false-positive or false-negative results. Soft-tissue attenuation, mainly from the diaphragm and the breast, is a major cause of artifacts. ^5–7,14 Interpreting physicians who are aware of these problems may try to “read around” these artifacts, which can lead to underestimation of existing disease. Gated imaging provides information about wall motion and thickening of the myocardial segments, as well as ventricular function and volumes. The rationale behind its use in differentiating the defects due to soft-tissue attenuation from true perfusion defects is that areas with nonviable myocardium would not move or thicken normally. ⁸ Recently, Smanio and co-authors ⁹ showed that gating helps to significantly decrease the number of borderline interpretations, increasing the numbers in the definitely normal and definitely abnormal categories. In their study, ⁹ the percentage of borderline interpretations before the addition of gating was about 30%. In our opinion, however, the interpreting physician in everyday practice, although not always entirely confident of his or her impression, is generally forced to commit to a diagnosis and decrease the number of borderline interpretations as much as possible. In this study, we evaluated the impact of routine gating of every MPI study on the interpretation of such studies in the day-to-day routine of a busy laboratory. To our surprise, the number of studies in which the results were changed with the addition of gating, either from abnormal to normal or vice versa, was very low (3.7%). It is known that small areas of scarred myocardium may have normal wall motion when limited to the subendocardium or when dragged/pulled along by adjacent normal tissue. Therefore, the assumption that normal wall motion excludes perfusion abnormalities may result in false-negative results in the diagnosis of coronary artery disease. The prognosis of such patients may still be good, but this is a topic for further investigation.

The percentage of studies in which gating eliminated the borderline interpretations was very small (1.1%). In a larger number of studies, the confidence level was increased but with no impact on the final interpretation. The difference between our study and that of Smanio's group ⁹ is that in the majority of our cases, the interpreting physicians, although not always entirely confident, committed themselves to a definite diagnosis before evaluating the gating. With few exceptions, these diagnoses remained unchanged after the addition of gated information. When we analyzed the effect of gating on interpretation and confidence levels in men and women separately, we found the advantage of gating to be more pronounced in men than in women. This result may be explained by the diaphragmatic attenuation or the effect of a wider chest, ^5,7 each of which is more common in men.

Therefore, our study demonstrates that the yield of routine gating of every MPI study for the differentiation between true defects and soft tissue attenuation is low for readers with long experience in interpretation of myocardial perfusion imaging. Although gating frequently increases the confidence level, if used routinely in every MPI study, it seldom leads to a change in interpretation. Experienced readers will be as accurate in their interpretations in the majority of cases by using only the static tomographic slices and committing themselves to their expert opinions. Further studies should be performed to evaluate the impact of gating in readers with limited experience in myocardial perfusion imaging interpretation. Each physician should also evaluate separately the impact of gating in his or her interpretation. It is likely that each physician for a certain period of time or for a certain number of cases needs gating to avoid borderline interpretations. Eventually, however, after a period of time that may vary from physician to physician, a level of expertise is reached at which gating is no longer needed.

Obviously, gating provides functional information as well as information about ventricular volumes not available in static perfusion images. It is therefore important to identify the specific subgroups of patients in which it offers an incremental value in diagnosis or prognosis.

Conclusions

Our study shows that the yield of routine gating of every MPI study for differentiation between true defects and soft-tissue attenuation is low for physicians experienced in the interpretation of myocardial perfusion imaging. Although gating information frequently increases confidence in the interpretation, it rarely leads to a change in the interpretation. It is therefore important to identify the specific subgroups of patients in whom gating offers an incremental value in diagnosis or prognosis.

Acknowledgment

The authors gratefully acknowledge S. Leticia Alanis-Williams, CNMT, for her assistance with the acquisition of myocardial perfusion and gating data.