Author + information
- Received May 13, 2015
- Revision received September 2, 2015
- Accepted September 3, 2015
- Published online June 1, 2016.
- Tadaki Nakahara, MD, PhD∗ (, )
- Yu Iwabuchi, MD and
- Koji Murakami, MD
- ↵∗Reprint requests and correspondence:
Dr. Tadaki Nakahara, Department of Diagnostic Radiology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan.
Objectives The aim of this study was to develop a display method to describe the fusion data of myocardial perfusion single-photon emission computed tomography (SPECT) and coronary computed tomography angiography into a single image that we call fusion-based bull’s eye (FBE). This study sought to show its generating process and evaluate its diagnostic performance.
Background Three-dimensional (3D) display is mostly used when reviewing SPECT/coronary computed tomography angiography fusion images, although multidirection interpretation is required to sweep the entire heart. Bull’s eye display of the fusion image will be useful in clinical practice.
Methods FBE images were generated from the 3D fusion data by determining a cardiac axis, adding a cylindrical object around the aortic root, obtaining a panoramic image from circumferential data of the 3D images, and converting it into a polar coordinate display image. The diagnostic performances of SPECT, the conventional 3D fusion, and FBE as to the presence of hemodynamically relevant coronary vessels were compared in 39 patients with abnormal SPECT findings.
Results The 3D fusion and FBE images were successfully obtained in all patients. Of an evaluated 105 coronary segments in 35 patients without coronary artery bypass grafting, SPECT showed 17 segments (16%) equivocal to determine hemodynamically relevant coronary vessels. FBE corrected the diagnoses of 5 segments, in which SPECT was false-negative in 2 or false-positive in 3, with only 2 equivocal segments (p = 0.0017). FBE also revealed 4 culprit lesions in all 4 patients with coronary artery bypass grafting. There were no discordances between FBE and the 3D fusion.
Conclusions FBE had the same capacity as the 3D fusion to solve equivocal SPECT findings or correct the diagnoses in 24 of 109 (22%) coronary segments for culprit lesion detection. Although FBE requires manual generation process at present, it facilitates evaluation of myocardial perfusion and coronary anatomy with only 1 image.
Myocardial perfusion single-photon emission computed tomography (SPECT) and coronary computed tomography angiography (CTA) are distinct diagnostic imaging modalities providing functional and anatomic information, respectively. SPECT studies have shown solid evidence for prognostic assessment based on the extent and severity of myocardial perfusion abnormalities in patients with coronary artery disease (1–4). However, coronary CTA has high sensitivity and negative predictive value for detecting coronary artery stenoses, and coronary CTA gains wider acceptance of its daily use because of advanced technologies with which higher quality images can be obtained with less amount of radiation (5,6). However, according to a recent large-scale study, coronary CTA failed to demonstrate prognostic benefit for evaluation of symptomatic patients with suspected coronary artery disease as compared with functional testing (7). The results are partly explained by the previous studies demonstrating that only structural information of coronary artery cannot predict hemodynamically significant stenoses that are linked to subsequent cardiac events (4,8–10).
SPECT–coronary CTA hybrid imaging may be one of the forms that routine myocardial perfusion imaging will take (11) because image fusion significantly improves detection of hemodynamically significant coronary lesions (11–13). Expanded availability of image fusion software facilitates the clinical use of SPECT–coronary CTA cardiac fusion using coronary CTA from external sources (14,15). Three-dimensional (3D) display is commonly used to exhibit this type of image fusion, although it sometimes requires image browsers supporting movie display for multidirection interpretation to avoid overlooking the potential disease. In this regard, under the circumstance of availability of both SPECT and coronary CTA, it would be better to display data on a single image like bull’s eye for SPECT.
Here, we developed a display method to describe the SPECT–coronary CTA data into a single image that we call “fusion-based bull’s eye” (FBE). As far as we searched the PubMed biomedical database (US National Library of Medicine, Bethesda, Maryland), no articles related to “bull’s eye” and “SPECT” and “CT” or “CTA” or “coronary CTA” have been published through April 1, 2015. The aims of this study were to show the process of FBE generation and evaluate its diagnostic performance.
Materials and Methods
Patients and imaging protocol
From January 2014 to August 2014, exercise (n = 344) or dipyridamole (n = 330) stress/rest SPECT was performed and 65 patients had abnormal SPECT findings. SPECT images were acquired 15 min after patients were injected with 111 MBq of thallium-201 chloride at the time of peak stress, followed by low-dose CT scanning for attenuation correction using SPECT/CT scanner (Discovery NM/CT 670pro, GE Healthcare, Waukesha, Wisconsin). SPECT data were collected in 60 views in steps of 6° with each detector rotating 180° (360° acquisition). Scan duration was 10 min, 12.5 min, and 15 min for patients weighing <60 kg, 60 to 70 kg, ≥70 kg, respectively. The matrix size for data acquisition and image reconstruction was 3.4 mm (128 × 128). Reconstructed SPECT images with 3.4-mm slice thickness were obtained with a Butterworth filter (order, 10; cutoff frequency, 0.33 cycles/cm). Unfortunately, the SPECT/CT scanner has an only 16-detector row CT, making it difficult to perform coronary CT angiography. Attenuation and scatter correction were applied.
Among 65 patients with abnormal SPECT findings, such as stress-induced ischemia and myocardial infarction, 39 underwent coronary CTA with the time interval of not more than 8 months. Patient demographics are shown in Table 1. Coronary CTA was performed using a 64-detector row CT (LightSpeed VCT, GE Healthcare) or a 320-detector row CT (Aquilion ONE, Toshiba Medical Systems, Otawara-shi, Japan). Dose of oral metoprolol, which was administered 1 h before CT scanning, was adjusted for patient’s heart rate not to exceed 65 beats/min. Then, coronary CTA data with administration of glycerol trinitrate (0.3 mg) were obtained before and after intravenous injection of 0.7 ml/kg of iodine contrast material (Iopamiron 370, 370 mgI/ml Iopamidol; Bayer Healthcare, Tokyo, Japan). Data were reconstructed with 0.625-mm slice thickness and no overlap for the 64-detector and 0.5-mm slice thickness and 0.25-mm overlap for the 320-detector row CT. Both SPECT with low-dose CT for attenuation correction and coronary CTA data acquisitions were approved by the institutional review board and all subjects consented in writing.
Conventional 3D fusion image generation
Conventional fusion process of SPECT and coronary CTA images was described elsewhere (11). Briefly, dedicated SPECT-CT fusion software (CardIQ Fusion SPECT, GE Healthcare) installed on an image browser (AW server 2, GE Healthcare) was used to generate 3D volume rendering coronary CTA images and extract the volume data of the coronary vessels and the left ventricle. Then, thallium-201 SPECT data in color were semi-automatically projected onto the surface of the left ventricle of the 3D coronary CTA images after aligning myocardial perfusion data in SPECT with the ventricular wall in coronary CTA (11).
FBE image generation
FBE is a modified display form of the original 3D SPECT–coronary CTA fusion (Figure 1A). First, a cardiac transparency tool in the fusion software is used to discriminate intraventricular contrast media and ventricular wall on the 3D coronary CTA images and then a cardiac axis is determined by rotating the transparent data (Figure 1B). Apical side of the cardiac axis is set at the tip of left ventricle chamber in the 3D images. The other side of the cardiac axis is set at the centroid of an elliptical plane of the left ventricular base. Coronary anatomy is colored in black. Then, an opaque white cylindrical object around the aortic root is added to conceal the far side of the coronary anatomy (Figure 1C). The object can be extracted from the original chest CT volume data. In addition, branches not supplying the left ventricle, such as acute marginal arteries, are manually eliminated. By rotating the fusion images along the cardiac axis, 540-direction circumferential images per 360 degrees are obtained and then digitized to 1,024 × 1,024 matrix using a continuous photographing function (Figure 1D). All the previously mentioned procedures can easily be performed by using the image workstation (AW server 2, GE Healthcare).
These data were transported into a commercially available personal computer. With commercially available image-editing software (Adobe Photoshop CC, Adobe Systems, San Jose, California), a thin rectangular image with a width of 1 pixel (0.15 to 0.20 mm) is extracted from the central axis on each of the 540 images, and a panoramic image is created by connecting them (Figure 2A). An opaque cylindrical object (Figure 1C) is effective to avoid inappropriate projection of coronary artery vessels onto the panoramic image. Finally, the panoramic image is trimmed into a square image and then converted into a polar coordinate display image with the software (Figure 2B).
Image interpretation was performed by 1 specialist who has a 17-year experience in both diagnostic radiology and nuclear medicine. First, SPECT images of the 39 patients were interpreted and coronary arteries responsible for abnormal SPECT findings (i.e., hemodynamically relevant coronary vessels) were determined without prior knowledge of coronary CTA findings. Because there is a variation of coronary anatomy, the interpretation was based on the most standard locations of left anterior descending artery, left circumferential artery, and right coronary artery (16). In this study, high lateral branch is considered being included in left circumferential artery. Because this study focused on determination of culprit lesion detection, abnormal findings were defined to be positive if they were located within areas of typical coronary artery territories (Figure 3). If reduced perfusion areas were observed from one coronary artery territory to the marginal zone of another coronary artery territory or located in the boundaries between these territories, the findings were defined to be equivocal.
Afterward, 3D SPECT–coronary CTA fusion images were interpreted with knowledge of presence or absence of coronary arteries with a diameter stenosis >50%. If abnormal perfusion areas and the territories of significantly stenotic coronary arteries match closely on fusion images, the lesions are considered as hemodynamically relevant coronary vessels (true-positive). If abnormal perfusion areas are incongruous with coronary artery territories or cannot be explained by coronary artery luminal structures (i.e., no sign of coronary artery stenosis) or other tests including invasive angiography, electrocardiography, and ultrasound cardiography, the lesions are unlikely to be hemodynamically relevant (false-positive).
After an interval of 1 month, FBE images were interpreted in the same manner as the conventional 3D fusion images. The order of the 39 FBE images was randomly changed from that of the 3D fusion images beforehand.
The generalized McNemar test was used to determine if there are diagnostic differences between each pair of the 3 methods: SPECT alone, the conventional 3D fusion, and FBE. The analysis was performed using SAS 9.2 software (SAS Institute, Cary, North Carolina).
Table 2 shows a comparison of interpretation results of hemodynamically relevant coronary vessels among SPECT alone, the conventional 3D fusion, and FBE. SPECT showed 51 reduced perfusion areas in 35 patients without coronary artery bypass grafting (CABG) and 4 areas in 4 patients with CABG. There were 17 equivocal results among the 51 areas in patients without CABG (16%); 15 abnormal perfusion areas were extended from 1 coronary artery territory to the marginal zone of another coronary artery territory and 2 areas were localized in the boundaries between the territories of coronary vessels. Especially, there were 9 reduced perfusion areas that were difficult to determine whether culprit vessels were left circumflex coronary artery only, right coronary artery only, or both (Figure 4). Regarding the 4 abnormal perfusion areas in 4 patients with CABG, it was impossible to determine culprit vessels with difficulty of discriminating native and graft vessels (Figure 5).
Of the 17 equivocal findings with SPECT alone, FBE solved 15 findings in terms of culprit vessel determination in which 7 were clearly indicative of culprit lesions based on fusion images (positive) and 8 were not (negative). Among them, 4 areas tuned out to be caused by left dominant circulation. In addition, FBE modified the diagnoses of 5 segments in which SPECT was false-negative in 2 or false-positive in 3 segments. However, 2 segments that remained equivocal even after reviewing FBE images turned out to be noncompaction of the left ventricle in 1 and complete left bundle branch block in 1. As a result, there was a significant diagnostic difference between SPECT alone and FBE (p = 0.0017) (Table 3).
FBE successfully detected 4 culprit lesions in all 4 patients with CABG. If the 4 areas were assigned to equivocal for SPECT alone and positive for FBE, the diagnostic difference was more significant (p = 0.0002).
There were no discordances between the conventional 3D fusion and FBE. Therefore, statistical analysis could not be further performed.
Little has been seen in the literature as to application of bull’s eye map to hybrid imaging. Our method requires many procedures to accomplish the generation of FBE images, although each step is easy to understand and, more importantly, the original 3D fusion data were not changed or processed in all steps. In fact, there was 100% concordance achieved in diagnostic performance of FBE and the original 3D fusion. Recently, Kirişli et al. (17,18) developed unique fusion software and applied it to a clinical study. In their method, bull’s eye map seems to be created in a similar way; 3D ellipsoid myocardium is unfolded and reformatted into a 2-dimensional circular plane (17). However, their clinical study excluded patients with CABG and it is therefore unknown if their method could be applied to patients with CABG. By contrast, our method can be used for any type of patients because it allows coronary artery structures to be kept unchanged when the fusion data are converted onto FBE images. As shown in Figure 5, fusion imaging would be significantly effective in patients with CABG because of the complexity of coronary anatomy.
There may be several limitations of fusion image generation in clinical practice. First, a series of our procedures is time-consuming at present. However, the fully processed 3D volume rendering data of the coronary CTA, which should originally be prepared as a CT examination, can be used to make FBE and the 3D SPECT–coronary CTA fusion images. We are developing the automated system for generating FBE images. We believe that our method will gain wide acceptance if it is automated. Second, the current study did not include patients with no sign of abnormal SPECT findings because our retrospective study did not focus on comparison of SPECT and coronary CTA in terms of diagnostic accuracy. It remains unknown whether fusion imaging can change the interpretation of normal-appearing SPECT images in patients with multivessel or left main coronary artery disease (19,20). Third, diagnostic performance of FBE was evaluated by only 1 radiologist. Unfortunately, there were very few who could interpret both myocardial perfusion SPECT and coronary CTA in our institution. The participation of less experienced radiologists in this study was not proactively considered because interpretation of SPECT alone requires more considerable training and experience than that of coronary CTA. Therefore, the diagnostic benefit of SPECT alone should have been underestimated. Finally, it is difficult to quantitatively assess the extent of hypoperfused areas on FBE images as a result of the significant warping of functional and anatomic data.
We developed a display method to describe the fusion data into a single image that we call “fusion-based bull’s eye.” The diagnostic performance of FBE for culprit lesion detection was comparable with that of the original 3D fusion. Although FBE generation is time-consuming at present, the generation process is simple and understandable. In addition, FBE can be applied to patients with CABG. Under the circumstance of availability of both SPECT and coronary CTA, FBE will be one of the choices to review anatomic and functional information at the same time.
COMPETENCY IN MEDICAL KNOWLEDGE: Fusion imaging of myocardial perfusion SPECT and coronary CTA, which improves detection of hemodynamically significant coronary vessels, has been 3-dimensionally viewed. Display method of exhibiting such fusion image data into a single polar plot image was developed and termed “fusion-based bull’s eye.” The method is intended to generate a fused polar image like a conventional bull’s eye image combined with coronary anatomy, such that nuclear medicine and referring physicians can intuitively understand the correlation between myocardial perfusion and coronary vessels in an individual patient. Fusion-based bull’s eye showed the same capacity of original 3D fusion to solve equivocal SPECT findings or correct the diagnoses even in patients with CABG.
TRANSLATIONAL OUTLOOK: The method still requires manual generation process. Further improvements, such as automated generation process, is warranted to apply fusion-based bull’s eye in routine clinical practice.
The authors thank colleagues at the Departments of Diagnostic Radiology and Cardiology involved in myocardial perfusion scan and coronary computed tomography angiographic scan for this study.
All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery bypass grafting
- computed tomography angiography
- fusion-based bull’s eye
- single-photon emission computed tomography
- Received May 13, 2015.
- Revision received September 2, 2015.
- Accepted September 3, 2015.
- American College of Cardiology Foundation
- Hachamovitch R.,
- Berman D.S.,
- Shaw L.J.,
- et al.
- Iskandar S.,
- Iskandrian A.E.
- Sharir T.,
- Germano G.,
- Kang X.,
- et al.
- Shaw L.J.,
- Berman D.S.,
- Maron D.J.,
- et al.
- Stehli J.,
- Fuchs T.A.,
- Bull S.,
- et al.
- Pijls N.H.J.,
- van Schaardenburgh P.,
- Manoharan G.,
- et al.
- Tonino P.A.L.,
- Fearon W.F.,
- De Bruyne B.,
- et al.
- Rispler S.,
- Keidar Z.,
- Ghersin E.,
- et al.
- Gaemperli O.,
- Schepis T.,
- Valenta I.,
- et al.
- Slomka P.J.,
- Cheng V.Y.,
- Dey D.,
- et al.
- Cerqueira M.D.,
- Weissman N.J.,
- Dilsizian V.,
- et al.
- Kirişli H.A.,
- Gupta V.,
- Shahzad R.,
- et al.
- Ragosta M.,
- Bishop A.H.,
- Lipson L.C.,
- et al.