Author + information
- Received September 19, 2007
- Revision received November 26, 2007
- Accepted December 2, 2007
- Published online March 1, 2008.
- Tali Sharir, MD⁎,⁎ (, )
- Simona Ben-Haim, MD, DSc†,
- Konstantine Merzon, MD⁎,
- Vitali Prochorov, MD⁎,
- Dalia Dickman, PhD‡,
- Shlomo Ben-Haim, MD, DSc‡ and
- Daniel S. Berman, MD§
Reprint requests and correspondence:
Dr. Tali Sharir, Procardia, Maccabi Health Services, 156 Hayarkon St., Tel Aviv, Israel, 63451.
Objectives The purpose of this study was to compare myocardial perfusion imaging (MPI) with high-speed single-photon emission computed tomography (SPECT) with conventional SPECT imaging for the evaluation of myocardial perfusion in patients with known or suspected coronary artery disease.
Background A novel technology has been developed for high-speed SPECT MPI by employing a bank of independently controlled detector columns with large-hole tungsten collimators and multiple cadmium zinc telluride crystal arrays.
Methods A total of 44 patients (39 men) underwent same-day Tc-99m sestamibi stress/rest MPI. High-speed SPECT images were performed within 30 min after conventional SPECT. Stress and rest acquisition times were 16 and 12 min for conventional imaging and 4 and 2 min for high-speed SPECT, respectively. Myocardial counts/min (cpm) were calculated for both conventional SPECT and high-speed SPECT. Images were visually analyzed, and the summed stress score (SSS) and summed rest score (SRS) were calculated. Image quality and diagnostic confidence were qualitatively assessed.
Results High-speed SPECT SSS and SRS correlated linearly with conventional SPECT respective scores (r = 0.93, p < 0.0001 for SSS, and r = 0.93, p < 0.0001 for SRS). Image quality was rated good and higher in 17 (94%) cases for high-speed SPECT and 16 (89%) cases for conventional SPECT. Of the 44 patients studied, 36 (81.8%) and 35 (79.5%) were diagnosed definitely normal or abnormal by conventional and high-speed SPECT, respectively (p = NS). Myocardial count rate was significantly higher in high-speed versus conventional SPECT (384 × 10−3 ± 134 × 10−3 cpm/min vs. 47 × 10−3 ± 14 × 10−3 cpm/min, respectively, p < 0.0001) for stress and (962 × 10−3 ± 426 × 10−3 cpm/min vs. 136 × 10−3 ± 37 × 10−3 cpm/min, respectively, p < 0.001) for rest.
Conclusions High-speed SPECT provides fast MPI with high image quality and up to 8 times increased system sensitivity. The amount of perfusion abnormality visualized by high-speed SPECT is highly correlated to conventional SPECT, with an equivalent level of diagnostic confidence.
Single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) plays a major role in the diagnosis and prognostic evaluation of patients with coronary artery disease (CAD). Described 50 years ago (1), conventional gamma cameras for SPECT employ a large sodium iodide crystal and bulky photomultiplier tubes. Parallel-hole collimators provide positioning information but at a great cost in sensitivity with consequent prolonged imaging times. Implementation of dual detector cameras and computer developments has enhanced image quality (2–4) of the conventional Anger camera systems (conventional SPECT). Nevertheless, the trade-off between resolution and sensitivity dominates the performance of conventional gamma cameras and dictates the need for relatively large doses of radiopharmaceuticals and prolonged imaging time of 15 to 20 min.
A technology for high-speed perfusion imaging introduces a new design of both photon acquisition system as well as reconstruction algorithm (5). This system uses 9 pixilated solid-state detector columns, cadmium zinc telluride crystals (CZT), wide-angle tungsten collimators, region of interest (ROI)-centric scanning, and increased angular sampling. This technology has recently been shown to provide an 8- to 10-fold increase in sensitivity, coupled with a 2-fold improvement in spatial resolution (5,6), enabling a significant reduction in imaging time and dose of radio-isotopes.
The aim of this pilot clinical study was to assess the performance of a high-speed MPI in patients referred for evaluation of CAD, by a comparison with conventional SPECT imaging. We hypothesized that high-speed SPECT, acquired over a shorter time, will be comparable to conventional SPECT in detecting perfusion defects and will provide similar diagnostic confidence.
The study population consisted of 44 patients (39 men), referred for stress (exercise or dipyridamole)/rest Tc-99m sestamibi gated MPI at Procardia, Maccabi Healthcare Services, Tel Aviv, Israel. Patients were selected for the study when 1 of the following criteria was fulfilled: 1) low pretest likelihood of CAD; 2) moderate-to-high pretest likelihood of CAD in the absence of documented CAD; or 3) history of myocardial infarction. The study was approved by the local ethics committee. All patients signed an informed consent.
Exercise and dipyridamole protocols
Patients were instructed to discontinue beta-blocker drugs and calcium antagonists 48 h before testing and nitrates 24 h before testing. A symptom-limited treadmill exercise test was performed, with the Bruce protocol. Tc-99m sestamibi was injected at peak stress, and exercise continued at the same level for an additional 60 s and 2 min at 1 level lower. Horizontal or downsloping ST-segment depression ≥1 mm or upsloping ≥1.5 mm was considered positive for ischemia. For dipyridamole stress, patients were instructed to avoid caffeine-containing products for 24 h before the test. Dipyridamole (0.56 mg/kg) was injected over 4 min, and Tc-99m sestamibi was injected at 6 to 8 min after the beginning of dipyridamole injection. Whenever possible, patients performed a low-level treadmill exercise during dipyridamole testing.
Conventional SPECT acquisition protocol
Patients underwent stress/rest Tc-99m sestamibi gated SPECT. A low dose of 11 mCi Tc-99m sestamibi was injected at peak stress, and 8-frame gated SPECT imaging (100% acceptance window) was initiated 15 to 30 min later. A second dose of 28 mCi Tc-99m sestamibi was injected after 2 h, and 8-frame gated SPECT imaging was started 1 h later. Acquisitions were obtained with a 2-detector gamma camera (Axis, Philips Medical Systems, Cleveland, Ohio, or CardiaL, GE Healthcare, Haifa, Israel), 60 projections over 180° orbit, for 32 s/projection at post-stress and 24 s/projection at the resting acquisitions. The 8-projection sets were summed to generate an ungated data set used for perfusion assessment. Projection images were reconstructed with filtered back projection, and a 2-dimentional (2D) Butterworth post-filter was applied. No attenuation or scatter correction was applied.
High-speed SPECT: technology and acquisition protocol
The high-speed SPECT system (Spectrum-Dynamics, D-SPECT, Caesarea, Israel) uses 9 collimated, pixilated CZT detector columns, mounted vertically in 90° geometry (Fig. 1A). Each of the columns consists of 1,024 (16 × 64), 5-mm-thick CZT elements (2.46 × 2.46 mm). Square-hole tungsten collimators are fitted to each of the detectors, with the size of the collimator holes matching the dimensions the detector elements. The collimators are shorter than the conventional low-energy high-resolution collimators, yielding significantly better geometric efficiency. The resultant loss in spatial resolution is compensated by the use of a proprietary Broadview reconstruction algorithm (Chicago, Illinois), based on the maximum likelihood expectation maximization method (MLEM). Data are acquired by the detector columns rotating in synchrony, focusing on ROI (the heart) (Fig. 1) and saved in list mode.
Data were acquired in a semi-reclining position, with the left arm placed on top of the camera. A 30-s pre-scan acquisition was performed at the beginning of each imaging process, to identify the location of the heart within the chest and to set the angle limits of scanning for each detector (ROI-centric scanning). A 4-min post-stress acquisition (120 projection/detector, 2 s/projection) was performed approximately 30 min after post-stress conventional SPECT imaging, and a 2-min acquisition (1 s/projection) followed conventional SPECT resting imaging. Rest high-speed SPECT imaging was not obtained in 7 of the 44 patients.
Analysis of perfusion images
Stress and rest perfusion images of high-speed SPECT and conventional SPECT were semi-quantitatively scored with a 20-segment model of the left ventricle and a 5-point scale (0 = normal, 1 = equivocal, 2 = moderate, 3 = severe reduction of radioisotope uptake, and 4 = absence of detectable tracer uptake). Visual scoring of high-speed SPECT and conventional SPECT images was performed by an experienced nuclear cardiology physician (T.S.) on separate occasions, blinded to the scores given for the other imaging modality. The global summed stress score (SSS) and summed rest score (SRS) were calculated by adding the scores of the 20 segments in the stress and rest images, respectively.
Photon sensitivity, image quality, and diagnostic confidence
The sensitivity of high-speed SPECT and conventional SPECT were assessed as myocardial counts/min from an ROI on raw images, isolating the myocardium from surrounding structures. A 2D image was generated by averaging the projection images from the multiple high-speed SPECT detector heads, sharing the same viewing angle, onto a single plane from a virtual single head detector. Total counts of the entire image and the ROI for each projection were recorded and summed.
Conventional SPECT and high-speed SPECT images were scored for image quality with a 5-point scale: 1 = poor, 2 = fair, 3 = good, 4 = very good, and 5 = excellent.
Diagnostic confidence was graded for both imaging modalities being definitely normal, probably normal, equivocal, probably abnormal, and definitely abnormal.
Continuous variables (Table 1) are presented as mean ± SD. Correlation between SSS and SRS obtained by high-speed SPECT was evaluated by linear regression. Agreement between the 2 methods was assessed by Bland-Altman analysis (7). Comparison of myocardial count rate by conventional SPECT and high-speed SPECT was performed with paired Student t test. Image quality and diagnostic confidence for high-speed SPECT and conventional SPECT were compared with chi-square and Fisher exact tests. A p value < 0.05 was considered significant.
Table 1 summarizes clinical characteristics of the patient group. Mean age was 60 ± 11 years; 39 (88.6%) were men. Pre-scan likelihood of CAD, on the basis of age, gender, risk factors of CAD, symptoms, and history of CAD, was low in 10 (25%) patients and intermediate-to-high in 34 (75%) patients. Sixteen patients (36%) were asymptomatic. Of the 44 patients, 30 (68.2%) performed treadmill exercise, whereas 14 (31.8%) underwent dipyridamole stress combined with treadmill exercise, adjusted to the patient’s ability. Seven patients underwent invasive coronary angiography within 60 days of the perfusion study.
Correlation of perfusion scores
As shown in Figure 2A, there was excellent linear correlation between SSS by conventional SPECT versus SSS by high-speed SPECT (R = 0.91, p < 0.00001). The Bland-Altman analysis shows no correlation between conventional SPECT SSS minus high-speed SPECT SSS versus mean SSS (Fig. 2B). The mean difference between conventional SPECT and high-speed SPECT was small (1.27 ± 3.83), with all differences except for 2 located between the mean difference ± 2 SDs (−6.39 to 8.93).
Similarly, SRS by high-speed SPECT excellently correlated to SRS by conventional SPECT (R = 0.93, p < 0.00001) (Fig. 3A). The Bland-Altman analysis for the SRS showed no correlation between conventional SPECT SRS minus high-speed SPECT SRS versus mean SRS, with a very small mean difference of SRS values (0.89 ± 2.99) and most individual difference between the mean ± 2 SDs (−5.09 to 6.88) (Fig. 3B).
All 10 patients with low pre-scan likelihood of CAD had normal perfusion (SSS <3) by high-speed SPECT and by conventional SPECT. Figure 4 demonstrates high-speed SPECT and conventional SPECT images of a patient with low pre-scan likelihood, showing normal myocardial perfusion by the 2 imaging techniques.
Myocardial count rate, image quality, and diagnostic confidence
Myocardial count rate was significantly higher (7- to 8-fold) for high-speed SPECT compared with conventional SPECT at stress and rest (Fig. 5). Stress images were graded “good,” “very good,” or “excellent” in 42 (95.4%) of the 44 conventional SPECT images and 43 (97.7%) of high-speed SPECT images (p = 0.6). Of the 37 rest images, 36 (97.3%) for conventional SPECT and 35 (94.6%) for high-speed SPECT were graded “good” or higher (p = 0.6). Diagnostic confidence in image interpretation was also similar for the 2 imaging modalities (Fig. 6). Of the 44 patients, 36 (81.8%) and 35 (79.5%) had definitely normal or abnormal scan by conventional SPECT and high-speed SPECT, respectively (p = NS).
Correlation with coronary angiography
Tables 2 and 3 ⇓ ⇓ summarize high-speed SPECT, conventional SPECT, and angiographic results of the 7 patients who underwent coronary angiography within 60 days after the nuclear testing. A good correlation was observed between nuclear data by both nuclear imaging techniques compared with the angiographic data. Of the 7 patients, 6 had CAD by angiography, with coronary stenosis >50% of at least 1 of the major coronary arteries or their branches. One patient (#5) had normal coronary arteries. This patient had mild, partially reversible inferior wall perfusion defect by conventional SPECT but normal perfusion by high-speed SPECT. Patient #6 had 3-vessel disease by both conventional SPECT and high-speed SPECT, verified by angiography.
Figure 7 demonstrates conventional SPECT and high-speed SPECT perfusion images of a 57-year-old man (Patient #3 in Tables 2 and 3) with a history of anterior myocardial infarction and stent insertion at the middle portion of the left anterior descending coronary artery (LAD), referred for nuclear testing due to typical angina. Perfusion images of both techniques demonstrated severe, partially reversible defect at the mid-distal anterior, septal, and apical walls and a mild, reversible perfusion abnormality at the inferior wall. These findings are consistent with ischemia + infarct at the LAD territory and ischemia at the right coronary artery (RCA) distribution. Of note, high-speed SPECT images are characterized by better resolution with clearer edge definition and thinner ventricular walls. The inferior ischemia is better seen with high-speed SPECT images. The findings at coronary angiography, performed 34 days after nuclear testing, were complete occlusion of mid LAD with distal collateral filling, 90% stenosis of first septal branch, and 90% stenosis of mid RCA. The RCA was dilated with a stent.
This pilot study is the first clinical experience with high-speed myocardial perfusion SPECT imaging. The performance of high-speed SPECT was compared with conventional SPECT in detecting perfusion abnormalities, with a same-day, low-dose, high-dose, stress/rest Tc-99m sestamibi protocol, in 44 patients referred for evaluation of CAD at a single site. With the high-speed SPECT camera, images were acquired over 4 min for stress and 2 min for rest, compared with 16 min and 12 min, respectively, by conventional SPECT. The results of the study demonstrate significantly higher myocardial count rate by high-speed SPECT (7- to 8-fold), with comparable extent and severity of perfusion defects at stress and rest and similar diagnostic confidence.
New hardware and software developments
Two main approaches have been used to address the limitations of conventional Anger camera systems—improved photon acquisition systems and improved image reconstruction algorithms. Various cameras have been designed to improve spatial resolution and sensitivity as well as patient comfort, such as a pixilated detector system (Cardius XPO, Digirad Inc., Poway, California) with 1, 2, or 3 pixilated CsI (Tl) stationary detectors and photodiodes. Acquisition is performed for 7.5 min, with a rotating chair through an arc of 202.5°. This system reaches a reconstructed spatial resolution of 15.4 mm and a sensitivity of 234 cpm/μCi. A 3.5-min image acquisition with this system has been described with a modified reconstruction method and preliminary clinical results shown (8). The resolution achieved in this modification has not been reported. Another approach was presented by Funk et al. (9) with a 9-pinhole collimator system, enabling a 5-fold improved sensitivity with comparable resolution to a conventional gamma camera. There are no clinical data available yet validating this approach. Another cardiac SPECT system (CardiArc Inc., Lubbock, Texas) uses three 180° curved NaI (Tl) crystals and an array of photomultiplier tubes. A curved lead plate with 6 vertical slits is used for collimation, so that photons are detected by 6 separate sections of the crystals (6,10). Acquisition is performed by data sampling every 0.14° over an arc of 180°. The reconstructed spatial resolution of this system ranges from 3.6 mm at 82 mm from the slits to 7.8 mm at 337 mm. This system is not yet in clinical use at more than a single site.
Image quality has been improved by the development of iterative reconstruction algorithms, on the basis of MLEM and ordered set expectation maximization (OSEM). In addition, reconstruction software for resolution recovery, correcting for losses in spatial resolution due to line response function of the collimator, have been introduced (2–4). Various reconstruction algorithms (Wide Beam Reconstruction, 3D-Flash, Astonish, Evolution) as well as “motion frozen” processing of gated cardiac images improve image resolution and reduce blurring of the images by cardiac motion (11–13).
The high-speed SPECT camera employs a novel design of both photon detection system as well as reconstruction algorithm (14,15). These improvements enable significantly reduced imaging time or, alternatively, reduced radiation exposure, while providing higher resolution than the Anger camera approach. This system uses a pixilated solid-state detector, CZT. The 64,000-pixel detectors system has a unique ability to perform a patient-specific image acquisition, focusing the field of view on each patient’s heart. Also, the elimination of the large photo-multiplier tubes and thick detector crystals enables miniaturization of the camera dimensions, resulting in the development of an ergonomically optimized camera from both user and patient perspectives.
The present study demonstrates the feasibility of MPI with the high-speed SPECT technology, while significantly shortening imaging time; 4 min for stress images, and 2 min for resting images, with 30-s pre-scan acquisition for setting scanning limits of each detector. Acquisition was performed at a comfortable upright position, with the patient’s arms on the top of the camera. Patient motion during acquisition was not evaluated; however, the short imaging time and the comfortable patient and arm position is likely to reduce the occurrence of significant patient motion and motion artifacts.
The results of the study demonstrate excellent correlation between stress and rest perfusion abnormalities by high-speed SPECT versus conventional SPECT images (correlation coefficient of 0.93 for stress and rest). The Bland-Altman analysis demonstrated very low mean difference between conventional SPECT and high-speed SPECT for the SSS (1.27 ± 3.83) and SRS (0.89 ± 2.99), with most data points between the mean ± 2 SDs. Importantly, all patients with low likelihood of CAD had normal perfusion by high-speed SPECT, suggesting high specificity. Diagnostic confidence was similar for the 2 imaging methods, with about 80% of the studies categorized as definitely normal or abnormal.
The results of this study are based on a relatively small patient group from 1 site. Only 7 patients underwent coronary angiography after the nuclear testing; thus diagnostic accuracy was not assessed. Although list mode acquisition and acquiring of an R-wave marker is included in all high-speed SPECT studies, the present study did not evaluate the results of electrocardiographic gating of the acquired data, and contribution of regional wall motion assessment to the interpretation of perfusion images (16) was thus not assessed. Image interpretation was visual, without any quantitative assessment of myocardial perfusion, and used a 20-segment model of the left ventricle rather than the recently recommended 17-segment model. Future studies will present multicenter analysis of a larger patient group, comparing high-speed SPECT with conventional SPECT perfusion imaging, and will use quantitative assessment of myocardial perfusion, with normal limits specific for high-speed SPECT images.
High-speed MPI is a new technology, providing fast MPI with up to 8 times increased sensitivity and higher resolution than conventional SPECT. This pilot study demonstrated that the high-speed SPECT studies with 4-min stress and 2-min rest acquisitions resulted in high-quality images, perfusion abnormality highly correlated to conventional SPECT, and an equivalent level of diagnostic confidence. These findings establish the clinical feasibility of the high-speed MPI.
This study was funded by Spectrum Dynamics. Drs. Berman and Ben Haim own shares in Spectrum-Dynamics. Dr. Sharir is a consultant for Spectrum-Dynamics.
- Abbreviations and acronyms
- coronary artery disease
- cadmium zinc telluride
- left anterior descending coronary artery
- maximum likelihood expectation maximization method
- right coronary artery
- region of interest
- single-photon emission computed tomography
- summed rest score
- summed stress score
- Received September 19, 2007.
- Revision received November 26, 2007.
- Accepted December 2, 2007.
- American College of Cardiology Foundation
- Anger H.O.
- Patton J.,
- Sandler M.,
- Berman D.,
- et al.
- Maddahi J.,
- Mendez R.,
- Chuanyong B.,
- et al.
- Funk T.,
- Kirch D.L.,
- Koss J.E.,
- Botvinick E.,
- Hasegawa B.H.
- Madsen M.T.
- DePuey G.,
- Gadiraju R.,
- Anstett F.
- Slomka P.J.,
- Nishina H.,
- Berman D.S.,
- et al.
- Berman D.S.,
- Ziffer J.,
- Gambhir S.,
- et al.
- Sharir T.,
- Merzon K.,
- Prochorov V.,
- et al.
- Smanio P.E.,
- Watson D.D.,
- Segalla D.L.,
- Vinson E.L.,
- Smith W.H.,
- Beller G.A.