Introduction

Although stress myocardial perfusion scintigraphy (MPS) has a high sensitivity for detecting coronary artery disease (CAD) in patients with multi-vessel CAD,1,2,3,4,5,6,7,–8 it may be less potent in identifying the exact location of the diseased coronary arteries.9,10,11,12,–13 Several attenuation correction methods have been reported that show improved accuracy for identifying the number of narrowing coronary vessels among patients with multi-vessel disease.14,15,16,17,–18 Segmentation with scatter and photo-peak window data for attenuation correction (SSPAC) for single-photon emission computed tomography (SPECT) is a recently developed correction method that utilizes a patient-specific non-uniform attenuation coefficient map.19,20 We have previously shown that this method provides improved diagnostic accuracy for CAD without additional radiation exposure.21 In the present study, we investigated whether it is clinically feasible to estimate the number of diseased coronary artery vessels using the SSPAC method.

Methods

Study Patients

We examined 183 consecutive patients (121 men, 62 women; mean age, 72 ± 8 years) who had no prior myocardial infarction, decompensated congestive heart failure, idiopathic cardiomyopathy, atrial fibrillation, history of cardiac surgery, advanced atrioventricular block, severe chronic obstructive pulmonary disease, or bronchospasm, and who underwent 99mTc-tetrofosmin stress (physical or adenosine) and rest gated myocardial SPECT together with coronary angiography that was performed within 3 months of the SPECT study between 2012 and 2017. This study was approved by the Ethical Committee of the Osaka Medical College.

Image Acquisition

A 1-day stress and rest SPECT myocardial perfusion imaging (MPI) was performed in all patients. Eighty-nine patients (49%) performed an exercise-stress, and the remaining 94 patients (51%) were pharmacologically stressed (140 μg/kg·min of adenosine for 6 minutes). Twelve-lead electrocardiograms and blood pressure measurements were obtained at baseline and then every minute during exercise. Myocardial SPECT was performed approximately 30–45 minutes after administering a tracer injection of 296 MBq 99mTc-tetrofosmin using a triple-detector SPECT system (GCA-9300R; Toshiba Medical Systems, Tochigi, Japan) with low-energy high-resolution collimator. Four hours later, resting SPECT was performed approximately 1 hour after an injection of 600 MBq 99mTc-tetrofosmin. The gated SPECT data were acquired by the following parameters: 360° step-and-shoot rotation, 60 steps, 40 or 60 seconds per step, 128 × 128 matrix (1 pixel = 3.2 mm), 16 frames per cardiac cycle, and a beat acceptance window at 20% of the average RR interval calculated just before the beginning of data collection. The ungated projection data were obtained by summing the gated projection data.

Data Processing and Analysis

The SSPAC method was performed as described previously.21 The processes of SSPAC and non-correction images are shown in Figure 1. The SSPAC method required the acquisition of scatter window data (7% of the lower side of the emission window) in addition to photo-peak window data (140 keV ± 10%). A patient-specific attenuation correction map was obtained semi-automatically in a few minutes through SSPAC implemented in a GMS-7700b or Vitrea workstation (Toshiba Medical Systems, Tochigi, Japan). The non-correction images were reconstructed by performing filtered back projection from the projection data. The reconstructed short-axis images obtained by the SSPAC and non-correction methods were analyzed using a quantitative perfusion SPECT with gated SPECT software (QPS/QGS, Cedars-Sinai Medical Center, Los Angeles, CA, USA). Perfusion polar maps of 17 segments were generated using QPS software. SSPAC and non-correction databases were calculated using the QPS software. Ejection fraction, end-diastolic volume, and end-systolic volume were obtained using QGS analysis.22

Figure 1
figure 1

The processes for constructing segmentation with scatter and photo-peak window data for non-correction images and attenuation correction (SSPAC) images. A The non-correction images were reconstructed by performing filtered back projection from the projection data. B For SSPAC images, a triple-energy window (TEW) scatter correction was performed on the projection data. Those data were reconstructed by performing filtered back. The attenuation correction images were reconstructed by Chang’s iterative method using the attenuation coefficient map. ① The body and lung contours were determined from reconstructed scatter window data. ② The myocardium and the liver were determined from reconstructed photo-peak window data. ③ The body outlines, lung contours, myocardium, and liver were attached to models of the average mediastinum and thoracic spine, which were obtained by computed tomography. The attenuation coefficient value according to the influence of absorption was assigned to each organ. ④ A Gaussian filter was applied to the attenuation coefficient map to match to for the system resolution

Full size image

Evaluation of Perfusion

The SPECT images from the post-stress and resting studies were displayed on the PC monitor to evaluate myocardial perfusion, and two experienced observers who were blinded to the clinical data examined the images separately. Subsequently, the differences between the readers were resolved using consensus obtained by simultaneously viewing the SPECT images. We evaluated the diagnostic accuracy for the patient-based data and the vessel-based analysis.

A 17-segment model was used to semi-quantitatively evaluate the perfusion defects. Individual segments were scored as follows: 0, normal; 1, mildly reduced; 2, moderately reduced; 3, severely reduced; and 4, absent. The severity of reversible defects in myocardial perfusion was defined based on the summed difference scores (SDSs) derived from summed stress scores and summed rest scores for all the segments. The anterior, apical, and septal walls (1, 2, 7, 8, 13, 14, and 17 segments) were assigned to the left anterior descending artery (LAD), the lateral wall (5, 6, 11, 12, and 16 segments) was assigned to the left circumflex artery (LCx), and the left main trunk (LMT) disease, including both LAD and LCx territories, was assigned to LMT. The inferior and basal septal segments (3, 4, 9, 10, and 15 segments) were assigned to the right coronary artery (RCA).23 Abnormal MPS results were defined by a score of ≥2 of SDSs based on each coronary territory.

Coronary Angiography

Coronary angiography was performed using standard methods. Coronary stenosis was evaluated using multiple projections by experienced investigators who were unaware of the clinical data including SPECT data. Quantitative coronary angiography (QCA) was performed using QCA-CMS software (MEDIS medical imaging systems, Leiden, The Netherlands). A stenosis was considered significant when >50% lumen narrowing was present.24,25

Statistical Analysis

The data are presented as mean ± standard deviation (SD). Paired t-tests were used to compare the differences in the paired continuous data, and the χ2 or Fisher’s exact tests were used to compare the differences for the paired discrete data, as appropriate. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) in indices were calculated as predictors of significant CAD. The receiver operating characteristic (ROC) curves were generated by calculating the sensitivity and specificity for CAD. JMP 11.0 (SAS Institute, Cary, NC, USA) was used to perform all statistical analyses. A P value <.05 was considered to be significant.

Results

Clinical Characteristics

Using SSPAC method, attenuation coefficient maps could be created for 179 (98%) of the 183 study patients. The characteristics of these 179 patients are described in Table 1. Coronary angiography showed no significant stenosis in the coronary artery of 69 patients, showed single-vessel disease in 75 patients, and multi-vessel disease in 35 patients. Among the patients with multi-vessel disease, eight patients (23%) showed transient ischemic dilatation of >1.12, which was defined as an abnormal ratio.26,27

Table 1 Patient characteristics
Full size table

Diagnostic Accuracy of the SSPAC Method

The patient-based diagnostic ability of SSPAC and non-correction images for the detection of CAD is described in Table 2 (AUC 0.81 vs 0.75; P = .059), and vessel-based diagnostic ability is shown in Table 3 (AUC 0.83 vs 0.72; P < .05).

Table 2 Patient-based diagnostic ability of SSPAC and non-correction images in the detection of coronary artery disease
Full size table
Table 3 Vessel-based diagnostic ability of SSPAC and non-correction images in the detection of coronary artery disease in total 537 vessels from 179 patients
Full size table

Non-correction and SSPAC Image for Multi-vessel Disease

Among 35 patients with multi-vessel disease, the accuracy for diagnosing CAD in individual coronary territories is shown in Table 4. Even in patients with multi-vessel disease, the SSPAC method showed significantly higher sensitivity and NPV than the non-correction method. The respective proportion of number of stenosed coronary artery vessels as described at CAG, non-correction, and SSPAC method images were shown in Figure 2. Representative case of the SSPAC and non-correction images for the patient with multi-vessel disease is shown in Figure 3.

Table 4 Diagnostic ability of non-correction and SSPAC images in the detection of coronary artery disease based on the territory of vessels in 38 patients with multi-vessel disease
Full size table
Figure 2
figure 2

Respective proportion of the number of stenosed coronary artery vessels as described at coronary angiography (CAG), myocardial perfusion image; non-correction (NC) and SSPAC images

Full size image
Figure 3
figure 3

A 68-year-old man with significant stenosis in LAD and RCA. The non-correction myocardial perfusion images (A, C) reveal reversible defect in only the inferior wall, and images after attenuation correction using the SSPAC method reveal reversible defects in the anterior and inferior walls (B, D); however, apical thinning is more pronounced

Full size image

Discussion

This study demonstrated that MPS using the SSPAC method has similar specificity but higher sensitivity than the non-correction method in detecting significant coronary artery stenosis lesions, even in patients with multi-vessel disease.

The SSPAC method uses information which merges an atlas-based estimate of the bone structures, a segmented body and lung outline from a scatter window image, and myocardial and liver structure from photo-peak image, and soft tissues. The SSPAC method has the advantages of removing the risk of additional radiation exposure and minimizing the deleterious effect of respiratory motion while adding only a few minutes of analysis to construct the map. It has been reported that myocardial perfusion counts become more consistent with the use of SSPAC in normal subjects,20 and we also reported that the SSPAC method provides better diagnostic accuracy for the detection of CAD, including myocardial infarction using the summed stress score.20 In the present study, we selected patients who had no prior myocardial infarction; we diagnosed limited ischemic lesions using SDS scores. The SSPAC method also improved the precise per-vessel diagnosis in patients with CAD, and even in those with multi-vessel disease. Such improvement was reflected more clearly in the increased sensitivity in the LAD and RCA territory. We think that the following reasons may explain this: the myocardial perfusion became homogeneous in rest image, and scatter contribution to the apparent uptake in the myocardium was significantly different at rest and stress in their patients; this produces the change in SDS in the SSPAC method. The attenuation in the inferior region was corrected by SSPAC method and the counts in the other regions were corrected relatively. We also propose its combined with the ability to perform scatter correction to improve the contrast in the MPS image.

The sensitivity of stress MPS for detection of the presence of multi-vessel disease is limited.1,5,6,7,–8 The possible reasons for this include balanced perfusion abnormality, where the absence of a normal reference segment limits sensitivity; early plateau of tracer uptake, which limits detection of borderline stenosis; and early stoppage of exercise as a result of signs or symptoms due to the severest lesion.13 Extensive clinical investigations have been performed for attenuation correction and it is now one of the recommended techniques for improving the image quality of MPS and diagnostic accuracy for detecting CAD.28,29,30,–31 Use of attenuation correction for detecting multi-vessel disease is still controversial. Ficaro et al. have reported increased diagnostic accuracy of attenuation-corrected SPECT for diagnosis of multi-vessel disease by simultaneous transmission/emission tomography or computed tomography.18 In contrast, Hendel et al. showed that among 96 patients with angiographically proven CAD and 88 subjects with low likelihood of CAD, attenuation correction resulted in a significantly higher normalcy rate (86% vs 96%, respectively). However, the sensitivity did not improve with attenuation correction, and for patients with RCA or multi-vessel disease, it was slightly worse in a multi-center trial.32

Patients with multi-vessel disease have poor prognosis without revascularization.33 Underestimation of the extent of ischemia on SPECT MPI is regarded as one of its most important limitations. We thought that it is important to detect multi-vessel disease by MPS. Furthermore, the SSPAC method is suitable for routine attenuation correction at our institute. It is easy to perform and usually requires several minutes or less to construct the patient-specific attenuation map without additional radiation exposure.

Study Limitations

This study has some limitations. First, we were unable to construct attenuation coefficient maps using the SSPAC method for some patients in this study. The primary reason was that it was not possible to detect the outline of the body or lung from scatter image. We thought that patients with the following condition may not be suitable for the SSPAC method: very large patients, cases for which arms must be positioned at the side of the body, lung lobectomy, excessive 99mTc-tetrofosmine lung uptake, patients with large gastric bubble. Second, our study involved a relatively small number of patients. We need to assess this method over a larger population, especially in patients with multi-vessel diseases, and evaluate its effectiveness according to differences in sex, body mass index, and other parameters. Third, we reconstructed SPECT images using the filtered back projection and Chang’s iterative method. We need to assess the accuracy of this method in comparison with the ordered subset expectation maximization method. Fourth, visual assessment is subjective and experience dependent. Fifth, in this study, we used both attenuation and scatter correction to construct SSPAC image. Both these methods may occasionally lead to miscorrection. Further, we did not assess the SSPAC method and scatter correction separately to know the mechanism of improved diagnosis accuracy in stenosed vessels. Finally, the diagnostic criteria for coronary stenosis include stenosis of the coronary arteries according to angiographic findings, not physiological changes, such as the functional flow reserve.34,35,36,–37

Conclusions

The SSPAC method of stress-induced SPECT may provide improved sensitivity in the presence of stenosed coronary artery vessels, even in patients with multi-vessel CAD.

New Knowledge Gained

The SSPAC method may be useful for diagnosing in the presence of stenosed coronary artery vessels.