European Heart Journal (2002) 23, 1714 1725 doi:10.1053/euhj.2002.3264, available online at http://www.idealibrary.com on Non-invasive coronary angiography with high resolution multidetector-row computed tomography Results in 102 patients A. F. Kopp 1, S. Schroeder 2, A. Kuettner 1, A. Baumbach 3, C. Georg 1, R. Kuzo 4, M. Heuschmid 1, B. Ohnesorge 1, K. R. Karsch 3 and C. D. Claussen 1 1 Department of Diagnostic Radiology, 2 Department of Internal Medicine, Division of Cardiology, Eberhard-Karls-University Tuebingen, Germany; 3 Department of Cardiology, Bristol Heart Institute, University of Bristol, U.K.; 4 Mayo Clinic, Jacksonville, Florida, U.S.A. Aims A new generation of multidetector-row CT (MDCT) scanners allows complete coronary coverage using retrospective ECG gating and 1 mm slices. The purpose of this study was to investigate the potential of high resolution MDCT angiography with retrospective gating for detection of coronary artery stenoses. Methods and Results A total of 102 patients underwent both conventional and MDCT coronary angiography. After intravenous injection of a non-ionic contrast medium the entire heart was scanned within a single breath hold using 1 mm slices. All MDCT data sets were reconstructed with retrospective gating at 20% to 80% in increments of 10% relative to the cardiac cycle. Two blinded independent reviewers analysed image quality for segments 1 4 (right coronary artery), 5 8 (left main, left anterior descending), and 11, 12 (left circumflex). These segments were evaluated for the presence or absence of significant ( 50%) stenoses. The results were compared with those of invasive coronary angiography in a blinded fashion. Overall sensitivity for the detection of significant stenoses ( 50%) were 0 86 (reader 1) and 0 93 (reader 2), specificity 0 96 (reader 1) and 0 97 (reader 2), negative predictive value 0 98 (reader 1) and 0 99 (reader 2). Conclusions High resolution MDCT angiography with retrospective gating permits the non-invasive detection of coronary artery stenoses with high accuracy if image quality is optimized for each of the three major coronary arteries. (Eur Heart J, 2002; 23: 1714 1725, doi:10.1053/euhj.2002. 3264) 2002 The European Society of Cardiology. Published by Elsevier Science Ltd. All rights reserved. Key Words: Multidetector-row computed tomography (MDCT), coronary artery disease (CAD), computed tomography angiography (CTA), image reconstruction. Introduction Despite a multitude of different medical and interventional strategies to treat coronary artery disease (CAD), the natural course of CAD is a relentless progression. The current gold standard to assess the degree of stenosis is coronary angiography. In Germany alone, the total number of angiographic procedures rose by 45% from 1995 to 2000, while the fraction of interventional procedures remained almost constantly low at about 30% [1]. Although coronary angiography has become a safe procedure with only a small risk associated [2], the Revision submitted 5 April 2002, and accepted 10 April 2002. Correspondence: Andreas F. Kopp, MD, Department of Diagnostic Radiology, Eberhard-Karls-University Tuebingen, Hoppe-Seyler-Str. 3, 72076 Tuebingen, Germany. inconvenience for the patient as well as the economic burden have fueled the quest to find an alternative, non-invasive method to visualize and assess coronary arteries. Achenbach et al. introduced intravenous electron beam computed tomography (EBCT) coronary angiography as a non-invasive imaging modality for the diagnosis of coronary artery disease [3]. With EBCT imaging of the coronary arteries, data acquisition is triggered only prospectively with the patient s electrocardiogram by selecting one point in the cardiac cycle. This time point is fixed as soon as the data have been acquired. The timing for all cardiac vessels is identical with prospective triggering [3]. From conventional angiography [4,5] as well as from EBCT [6] it is well know that the motion pattern of each of the three major coronary arteries during the cardiac cycle is different. If 0195-668X/02/$35.00/0 2002 The European Society of Cardiology. Published by Elsevier Science Ltd. All rights reserved.
Non-invasive coronary angiography 1715 imaging is done only at one time point in the cardiac cycle the results can be optimal for only one of the three major coronary arteries [3]. This is probably the major reason for the high number of individual vessels that cannot be evaluated because of motion artifacts [6].Inhis study Achenbach et al. had to exclude 25% of all coronary segments because of inadequate image quality mainly due to atrial contraction during end-diastole [3]. This was especially true for the left circumflex and right coronary arteries. Furthermore, because of the restriction to axial, non-spiral scanning in ECG-synchronized cardiac investigations, acquisition of 3D volume images by using EBCT can only provide limited z-resolution of 3 mm slice thickness within a single breath-hold scan [7,8] ). Recently, mechanical multidetector-row CT (MDCT) systems with simultaneous acquisition of four slices and half-second scanner rotation have become available for general-purpose scanning [9]. Multidetector-row acquisition with these scanners allows for considerably faster coverage of the heart volume, compared to single-slice scanning. This increased scan speed can be used for retrospective gating together with 1 mm collimated slice widths. This allows coverage of the entire cardiac volume in one breath hold and provides improved spatial resolution for multidetector-row CT angiography (MDCTA) of the coronary arteries in comparison to EBCT [7,10]. The gantry rotation speed of 500 ms together with dedicated reconstruction algorithms make a temporal resolution of up to 125 ms feasible [7]. The technique of retrospective gating can be used to optimize image quality and detection of stenoses for each of the three major coronary arteries during the cardiac cycle. Finding the optimal time point with least motion for each vessel should reduce the number of vessels which are impossible to evaluate due to motion artifacts and increase diagnostic accuracy. The purpose of the present study was to compare retrospectively ECG-gated MDCT and quantitative coronary angiography (QCA) in 102 patients to assess the accuracy of this new method for the detection of stenoses and occlusions. Methods Subjects A total of 106 consecutive patients scheduled for conventional coronary angiography at our institution were enrolled between 16 October 1999, and June 2001. All patients underwent both conventional and MDCT angiography within 1 week. In all cases, MDCT angiography was performed before conventional angiography. The ethical committee of the hospital approved the trial, and all patients gave written informed consent. The study protocol complied with the Declaration of Helsinki. The patients mean age was 62 9 years. There were 79 men, and 27 women. Their average weight was 79 kg, but 15 patients weighed more than 100 kg. Exclusion criteria were: coronary stents, unstable clinical condition, contraindications to the administration of iodine-containing contrast agent. We also included patients not in sinus rhythm. Multidetector-row CT Acquisition of data/reconstruction of images All CT examinations were done on a Somatom Volume Zoom multidetector-row CT scanner (Siemens, Forchheim, Germany). The scan protocol used was 4 1 mm collimation, pitch 1 5 2 according to the minimal heart rate, 500 ms rotation time (T rot ), 120 kv, and 300 ma. Previous dose measurements with an Alderson phantom (Alderson Research Laboratories Inc., Long Island City, New York, U.S.A.) revealed an effective radiation dose of approx. 5 6 msv for male, and 6 7 msv for female patients using this scan protocol. To establish the scan delay time a test bolus of 15 cc contrast material and 20 cc saline chaser bolus was used. The circulation time was determined by measurements of CT density values in the ascending aorta. Imaging commenced at the circulation time plus 3 s. A bolus of 120 cc nonionic contrast (400 mg I. cc 1 ) was injected through an 18-gauge catheter into an antecubital vein at 4cc.s 1 followed by a 50 cc saline chaser bolus [11] with the use of a dual head powered injector (Medtron, Saarbrücken, Germany). All patients received 0 4 mg of nitroglycerin for dilation of the coronary epicardial arteries before imaging. Tomographic imaging commenced at the level of the aortic root above the coronary ostia. Patients were instructed to briefly hyperventilate just before the scanning and then hold their breath for the duration of the scanning. A continuous spiral scan was acquired with the ECG-signal recorded simultaneously. For reconstruction of the image data a newly developed algorithm was used [12]. A conventional single-sector algorithm was used below a heart rate of 70 beats. min 1. For heart rates above 70 beats. min 1 two sectors were used for reconstruction (Fig. 1). This adaptive approach provided a temporal resolution T rot /2 for imaging in the diastolic phase at moderate heart rates and higher temporal resolution up to T rot /4, i.e. 125 ms, for higher heart rates without a need for decreased spiral pitch and reduced z-resolution. To determine the optimal position of the image reconstruction window relative to the cardiac cycle seven different sets of reconstructions at 20%, 30%, 40%, 50%, 60%, 70%, and 80% of the cardiac cycle were performed for a small subvolume of each raw data file with the origin of the right and left coronary artery. The percentage denotes the center of the reconstruction window. From these images only those time points with least motion artifacts for each of the three major coronary arteries were chosen for complete reconstruction, i.e. a maximum number of three different complete reconstructions were used for evaluation. The optimal
1716 A. F. Kopp et al. Figure 1 Retrospectively ECG-gated four-slice spiral reconstruction with one-sector reconstruction (black bar) for heart rates <70 beats. min 1 and two-sector reconstruction (two grey bars are used for one reconstruction) for heart rates 70 beats. min 1. The broken lines indicate how the four detector rows travel along z-axis at a fixed speed (pitch). Using the adaptive approach gap-less volume reconstruction is possible with pitch 1 5 for all heart rates >40 beats. min 1. For heart rates 70 beats. min 1 the temporal resolution can be improved by using scan data from two cycles for reconstruction of an image ( segmented reconstruction ). The partial scan data set for reconstruction of one image then consists of two projection sectors from different heart cycles. With a gantry rotation of 500 ms (T rot ) this approach provides a temporal resolution of 250 ms (T rot /2) for imaging in the diastolic phase at moderate heart rates and higher temporal resolution up to 125 ms (T rot /4) for higher heart rates. Figure 2 Segmental anatomy of right coronary artery (RCA) (lateral view), and left coronary artery (right anterior oblique view) with left main (LM), left anterior descending (LAD), and left circumflex (L circumflex) according to American Heart Association (AHA). reconstruction windows were selected by consensus of two reviewers. The reconstructed slice width was 1 25 mm, image increment 0 6 mm. Evaluation of data The image data from the selected reconstructions with least motion artifacts were transferred to an off-line computer workstation (3D Virtuoso, SIEMENS, Erlangen) for 3D volume rendering postprocessing in a standardized fashion. The images were independently evaluated by two investigators without knowledge of the patients coronary angiogram. Cross-sectional precontrast and post-contrast images and sliding thin-slab maximum-intensity projections were available for analysis of calcifications and non-calcified plaques. For image analysis coronary segments as defined by the American Heart Association were used (Fig. 2) [13]. The AHA coronary segments were determined using the 3D volume rendering views. In a first step the readers had to assess the number of segments visible and adequate for evaluation for segments 1 4 (right coronary artery, RCA), 5 (left main, LM), 6 8 (left
Non-invasive coronary angiography 1717 anterior descending, LAD) and 11, 12 (left circumflex, LCX). In a second step they had to give the location (segment) and degree of stenosis. A stenosis was considered to be significant on the order of 50% diameter narrowing. Conventional X-ray coronary angiography Cardiac catheterisation and contrast enhanced X-ray coronary angiography were performed according to the standard techniques. The angiograms were evaluated by two cardiologists without knowledge of the MDCT coronary angiographic findings. To determine lesion severity quantitative analysis of coronary angiograms (QCA) was performed using QUANSAD software (ARRI, Munich, Germany). The catheter tip was used for calibration. Percent area stenosis and minimal lumen diameters were measured in projections showing highest lesion severity. Coronary artery segments were either classified as significantly obstructed (diameter reduction >50%) or as normal or not significantly obstructed (diameter reduction <50%). Statistical analysis was performed using SPSS (Statistical Package for Social Sciences, version 10.1). The interobserver variation was quantified by calculating the Kappa statistic. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for detection of significant stenoses by means of MDCTA were calculated. For multiple testing Multivariate Analysis of Variance (MANOVA) was performed. For multiple pairwise comparisons of means the Tukey-Test was used. A P value <0 05 was considered significant for all statistical evaluations. Results MDCTA of the coronary arteries could be performed in all patients without complications. Mean time interval between the angiogram and the MDCT study was 2 days. Four patients were excluded from the evaluation because of technical difficulties or failure of the data acquisition (patient did not hear breathing instructions; ECG not properly recorded; scan aborted; claustrophobic patient moved during the scan), i.e. the data of only 102 patients could be used for evaluation. The mean time needed for investigation was 12 3 min. Total mean breathholding scan time was 28 s 9 s and was well tolerated by all patients. Mean heart rate was 63 11 beats. min 1. The distribution of the image reconstruction window with least motion artifacts relative to the cardiac cycle is given in Table 1. The LAD was best visualized in mid-diastole at 60 70% for most patients. The time point for best image quality of the right coronary artery (segments 1 4) was significantly different (P<0 05) early in diastole at 40% (Fig. 3). The left circumflex artery (segments 11, 12) showed optimal image quality most frequently at 50%. The mean visualized length of the RCA, LAD and LCX is listed in Table 2 as mean number of visualized Table 1 Distribution of centre of image reconstruction window with least motion artifacts relative to the cardiac cycle % Cardiac cycle 20% 30% 40% 50% 60% 70% 80% RCA 2 15 43 25 12 4 1 LM, LAD 0 2 4 15 23 41 17 LCX 0 9 21 39 23 8 2 n=patients; RCA=right coronary artery; LM=left main; LAD=left anterior descending; LCX=left circumflex. Table 2 Average number (percentage) of segments that could be evaluated Vessel segments. The main reasons for impaired image quality of individual segments were extensive calcification [33], very small arterial lumen [31], cardiac motion [24], artifacts of respiration [25], or the inability to discriminate the vessel from adjacent contrast-filled structures [19], i.e. veins or ventricle. Among the 102 patients the conventional coronary angiogram showed disease of one vessel in 37, two vessels in 20, and three vessels in 25 patients. Twenty patients showed no significant lesion (Fig. 4). For segments 1 8, 11 and 12 conventional angio showed a total of 142 significant stenoses: 20 in segment 1, 19 in segment 2, 10 in segment 3, three in segment 4, one in segment 5, 29 in segment 6, 28 in segment 7, four in segment 8, 22 in segment 11, and six in segment 12 (Figs 5 8). The number of false and right positive and negative findings for detection of significant stenoses in MDCTA are listed in Table 3. If all segments were included in the evaluation for detection of significant stenoses reader no. 1 found a sensitivity of 86%, a specificity of 96%, a positive predictive value of 81%, and a negative predictive value of 98%. The sensitivity was higher for reader no. 2; however, specificity and negative predictive value showed no statistically significant difference (P<0 05). Kappa for overall interobserver variation was 0 75. Discussion Number of segments RCA (four segments: 1 4) 2 8 0 6 (70%) LM (one segment: 5) 1 0 0 (100%) LAD (three segments 6 8) 2 9 0 6 (87%) LCX (two segments 11, 12) 1 8 0 5 (90%) See Table 1 for explanation of abbreviations. A challenging application for cardiac CT imaging is the non-invasive assessment of the coronary arteries for diagnosis of CAD. Limited temporal resolution of current CT technology requires synchronization of the
1718 A. F. Kopp et al. Figure 3 (a, b) Sixty-year-old patient, mean heart rate 65 beats. min 1 ; image reconstruction for RCA (arrow) at (A) 60% and (B) 40%. At 60% the RCA is almost impossible to identify. However, at 40% even the ramus posterolateralis dexter (arrowhead) can be identified. image acquisition to the movement of the heart by using ECG-information to provide phase consistent data in phases of low cardiac motion. For sequential imaging, a prospective trigger is derived from the ECG-trace to initiate the CT-scan with a certain user selectable delay time after the R-wave. The delay time for scan acquisition after an R-wave is calculated from a given phase parameter (e.g. a percentage of the RR-interval time as delay after an R-wave) for each cardiac cycle individually based on a prospective estimation of the RRintervals. Usually, the delay is defined such that the scans are acquired during the diastolic phase of the heart cycle. From cineangiographic studies [4,5], however, it is known that each of the coronary arteries has a distinct motion pattern in the course of the cardiac cycle. Because of their position in the coronary groove, the right coronary and the left circumflex artery have more rapid diastolic motion than the left anterior descending. The motion is caused mainly by atrial contraction during end-diastole [3]. This corresponds to the fact that in most EBCT studies the results for the right coronary arteries in regard to motion artifacts and image quality were worse than for the LAD [3,14]. Recently Achenbach et al. analysed the pattern of coronary arterial movement using EBCT [6]. He confirmed the finding, that each of the three major coronary arteries has as different motion pattern during the cardiac cycle. Achenbach et al. pointed out that the timing of the short acquisition window of EBCT within the cardiac cycle might be even more crucial than the timing of a longer acquisition window, as the variation in cardiac motion during acquisition levels out with the increasing duration of the acquisition window [6]. The different motion pattern of the individual coronary vessels calls for an individual reconstruction for each vessel in regard to position in the cardiac cycle. This can only be obtained if the data set contains data from all phases of the cardiac cycle. EBCT imaging only allows for sequential prospectively triggered acquisition [3]. The user has to select the phase of reconstruction in advance without being able to adapt or optimize it afterwards. Only one phase can be selected for all three vessels, i.e. a compromise has to be made in regard to the optimal time point for the individual vessels. The only means to achieve a reconstruction adapted to the phase to the cardiac cycle is retrospective gating. In addition to the advantages of phase selective image reconstruction, ECG-gated spiral scanning provides continuous volume coverage and better spatial resolution in the patients longitudinal direction as images can be reconstructed with overlapping increment. A 3D volume image can be reconstructed with a voxel size of about 0 5 0 5 1 0 mm based on a scan with 1 mm slice collimation and reconstruction with submillimetre image increment [7]. This is important for visualization of the right and left circumflex coronary arteries which run perpendicular to the imaging plane. With EBCT these vessels are visualized with lower spatial resolution than the LAD, which is oriented parallel to the imaging plane [3]. Using this retrospective gating technique we have demonstrated high overall sensitivity and specificity of 0 86/0 93 (reader 1/reader 2) and 0 96/0 97 (reader 1/reader 2), respectively, and a high negative predictive value (0 98/0 99) for MDCT coronary angiography for the detection of significant stenoses (>50% reduction in diameter) in the proximal, middle and distal coronary arteries in 102 patients. To the best of our knowledge this is the first study in which CT angiography of the
Non-invasive coronary angiography 1719 Figure 4 Three-dimensional reconstruction of coronary MDCTA with volume rendering technique (a, b, d) and maximal intensity projection (MIP) (c). View from anterior face (a) shows left main and left anterior descending with diagonal branches. Lateral view depicts left anterior (short arrow) descending and left circumflex artery (arrow). MIP image of right coronary artery segments 1 3 (c) shows small calcified plaques without significant stenosis. View from interior face (d) depicts both the posterior descending coronary artery running in the interventricular groove, and the posterolateral branch of the right coronary artery. coronary arteries includes the distal segments in the evaluation. In addition, all vessels were included in the evaluation; no segment was excluded due to technical limitations. For EBCT, Achenbach et al. reported a sensitivity of 92% for the detection of high grade stenoses in the proximal and middle coronary arteries, when excluding vessels that could not be evaluated; however, when all vessels were included the sensitivity was only 69% for just the proximal and middle segments [3]. Our results indicate that image quality and diagnostic accuracy in CT angiography of the coronary arteries can be significantly improved by high-resolution multidetector-row CT with individual selection of
1720 A. F. Kopp et al. Figure 5 (a d). different time points for reconstruction during the cardiac cycle for each of the three major coronary arteries. This is confirmed by results from other groups obtained in smaller study populations for detection of coronaryartery stenoses using retrospectively gated MDCTA. Niemann et al. reported a sensitivity of 81%, a specificity of 97%, a positive predictive value of 81% and a negative predictive value of 97% for detection of stenoses >50% in 35 patients [15]. Achenbach et al. found a sensitivity of 91% and a specificity of 84% in 64 patients [16]. Positive predictive value was 59%, negative predictive value 98% for the detection of high-grade coronary artery stenoses [16]. This high negative predictive value was also confirmed in our study. Thus, based on these results one potential clinical application of MDCTA might be to rule out coronary artery stenoses in coronary arteries that would be potential targets of revascularization therapy. This mainly includes coronary arteries with a vessel diameter 2 mm, which is well above the spatial resolution of MDCTA. Stenoses in vessels with a diameter <2 mm rarely constitute targets for revascularization.
Non-invasive coronary angiography 1721 Figure 5 (e h). Figure 5 Male patient, age 55, with situs inversus and two-vessel disease, s/p balloon angioplasty of proximal RCA and prox. LAD; a 60% lesion of distal RCA was left untreated. (a) Anterior view (volume rendering mode) of thorax with ventral thorax wall cut away depicts situs inversus. (b) RAO projection of RCX and prominent marginal branch in volume rendering mode. No high-grade lesion can be readily appreciated. (c) Maximum intensity projection (MIP) of RCX clearly shows a lesion (arrow). The degree of stenosis is estimated to be 70%. (d) RAO cranial view of proximal LAD in volume rendering mode. The high-grade lesion was not clearly delineated in this view. (e) MIP-image clearly depicts the lesion of the proximal LAD. (f) MIP of RCA in left posterior oblique projection. The arrow depicts the distal RCA lesion. The black arrow shows the area of the former dilatation, no restenosis present. (g) LAO cranial view of LAD and RCX with conventional angio. The arrows depict the lesions in each of the vessels. (h) RAO view of RCA by conventional angiography. Note the progress of the former 60% lesion to a high-grade lesion and the absent restenosis of the proximal RCA.
1722 A. F. Kopp et al. Figure 6 Forty-nine-year-old male with CHD of the RCA. S/P posterolateral myocardial infarction occurred 4 months previously. Patient underwent balloon angioplasty of subtotal stenosis at the beginning of the descending part of the RCA. Patient came in for follow-up performed with both coronary CT angiography (a) and conventional angiography (b). Both conventional angiography and CT angiography clearly show the patency of the RCA at the level of the balloon angioplasty. There is only minor residual stenosis of approx. 10 20% (arrows). Figure 7 MDCT coronary angiography and conventional coronary angiography: Volume rendered image (a) depicts high grade stenosis (arrow in the LAD). This finding is confirmed at conventional angiography (b). Limitations A limitation of our study is that we did not look individually at each segment of the different vessels for optimization of the reconstruction window. The patterns of movement obviously are not equal throughout the complete arterial tree. One might even have to look at proximal, mid, and distal segments individually to
Non-invasive coronary angiography 1723 Figure 8 (a c) MDCT coronary angiography and conventional coronary angiography: Volume rendered image (a) depicts high grade stenosis (arrow) in the LAD. Sliding thin-slab maximum-intensity projection (slab thickness 8 mm) (b) in this location depicts eccentric non-calcified plaque. Invasive coronary angiography (c) confirms presence of stenosis (arrow; 90% diameter reduction). account for different motion patterns. In addition, an increment of 10% for the position of the image reconstruction window might be too long to detect subtle differences. Further measures have to be taken to reduce the number of unevaluable segments. This may be achieved through shorter acquisition times by multidetector-row scanners with more than four channels. This would also permit the use of thinner collimation to improve spatial resolution in the z-axis. Shorter gantry rotation ( 500 ms) would shorten the image acquisition window to further reduce motion artifacts. Despite the fact that the CT scan per se is completed within a couple of minutes the entire procedure including processing and evaluation of data is still very time-consuming. Reconstruction of CT data, selection of optimal reconstruction windows and 3D postprocessing for interactive viewing took more than 1 h
1724 A. F. Kopp et al. Table 3 Detection of significant stenoses ( 50%) in multidetector-row CT coronary angiography. Results for both readers (nos 1 and 2) True positive True negative False positive False negative Reader no. 1 Reader no. 2 Reader no. 1 Reader no. 2 Reader no. 1 Reader no. 2 Reader no. 1 Reader no. 2 Segment 1 19 20 79 81 3 3 1 0 Segment 2 18 18 78 80 4 2 1 1 Segment 3 4 6 85 90 6 4 3 2 Segment 4 1 3 43 51 0 0 1 0 Segment 5 1 1 101 101 0 2 0 0 Segment 6 28 29 70 71 4 5 0 0 Segment 7 29 27 64 69 9 3 0 1 Segment 8 2 3 75 80 3 2 1 1 Segment 11 20 22 75 77 2 9 1 0 Segment 12 1 3 73 73 8 3 5 2 Table 4 Diagnostic accuracy of multidetector-row CT coronary angiography for the detection of significant stenoses ( 50%) and occlusions. The results are listed if (a) only the assessable segments were included, (b) all segments were included, (c) only the proximal segments were included for calculation of sensitivity, specificity, PPV, and NPV Assessable segments All segments (1 8, 11, 12) Proximal segments only (1 2, 5, 6 7, 11) Reader no. 1 Reader no. 2 Reader no. 1 Reader no. 2 Reader no. 1 Reader no. 2 Sensitivity 0 90 0 95 0 86 0 93 0 97 0 99 Specificity 0 95 0 96 0 96 0 97 0 97 0 98 PPV 0 76 0 81 0 76 0 81 0 87 0 92 NPV 0 98 0 99 0 98 0 99 0 99 1 0 PPV=indicates positive predictive value; NPV=indicates negative predictive value. per patient even when performed by an experienced reviewer. For routine clinical application this could eventually be significantly shortened and facilitated by dedicated software tools. During ECG-gated spiral imaging of the heart, data are acquired with overlapping spiral pitch and continuous X-ray exposure. Thus, ECG-gated spiral acquisition requires a higher patient dose than ECGtriggered sequential acquisition for a comparable signal-to-noise ratio. When performing multiple reconstructions in different cardiac phases for optimal image quality of individual vessels, almost all spiral data are used for image reconstructions and only a minimal amount of data is omitted. To obtain the same diagnostic information multiple sequential acquisitions would have to be performed with repeated injections of contrast material. This would eventually result in the same or even higher X-ray exposure. However, developments are under way that allow for a reduction of X-ray exposure for ECG-gated spiral acquisition by prospectively ECG-controlled on-line modulation of the tube output [17]. Preliminary results indicate that by reducing the tube output during heart phases that are not likely to be targeted by the ECG-gated reconstruction, dose savings up to 40% are possible [17]. This technique combines the important benefits of ECG-gated spiral scanning with X-ray exposure that is comparable to ECG-triggered sequential acquisition. Conclusion Despite the fact that this new technique is only in the early stages of development the results obtained so far with MDCT for non-invasive coronary angiography are encouraging and justify further research to evaluate its applicability and clinical usefulness. Dedicated software tools to reduce the time needed for image postprocessing and evaluation, together with further technical refinements such as simultaneous acquisition of more than four rows and faster gantry rotation speed, might help to accelerate the clinical implementation of non-invasive CT coronary angiography. References [1] Mannebach H, Hamm C, Horstkotte D. 17th report of performance statistics of heart catheterization laboratories in Germany. Results of a combined survey by the Committee of Clinical Cardiology and the Interventional Cardiology (for ESC) and Angiology Working Groups of the German Society of Cardiology-Cardiovascular Research for the year 2000. Z Kardiol 2001; 90: 665 7. [2] Kwok BW, Lim TT. Cortical blindness following coronary angiography. Singapore Med J 2000; 41: 604 5. [3] Achenbach S, Moshage W, Ropers D, Nossen J, Daniel WG. Value of electron-beam computed tomography for the noninvasive detection of high-grade coronary-artery stenoses and occlusions. N Engl J Med 1998; 339: 1964 71.
Non-invasive coronary angiography 1725 [4] Paulin S. Coronary angiography: a technical, anatomic, and clinical study. Acta Radiol 1964; 233S: 1 215. [5] Potel MJ, Rubin J, MacKay SA, Aisen A, Al-Sadir J, Sayre RE. Methods for evaluation cardiac wall motion in three dimensions using bifurcations points of the coronary arterial tree. Invest Radiol 1983; 18: 47 57. [6] Achenbach S, Ropers D, Holle J, Muschiol G, Daniel WG, Moshage W. In-plane coronary arterial motion velocity: measurement with electron-beam CT. Radiology 2000; 216: 457 63. [7] Ohnesorge B, Flohr T, Kopp AF et al. Cardiac imaging by means of electrocardiographically gated multisection spiral CT: initial experience. Radiology 2000; 217: 564 71. [8] Ohnesorge B, Flohr T, Becker C et al. Herzbildgebung mit schneller, retrospektiv EKG-synchronisierter Mehrschichtspiral-CT. Radiologe 2000; 40: 111 17. [9] Klingenbeck-Regn K, Schaller S, Flohr T, Ohnesorge B, Kopp AF, Baum U. Subsecond multislice computed tomography: basics and applications. Eur J Radiol 1999; 31: 110 24. [10] Kopp AF, Ohnesorge B, Flohr T et al. Multidetektor CT des Herzens: Erste klinische Anwendung einer retrospektiv EKGgesteuerten Spirale mit optimierter zeitlicher und örtlicher Auflösung zur Darstellung der Herzkranzgefäße. Fortschr Röntgenstr 2000; 172: 1 7. [11] Kopp AF, Ohnesorge B, Flohr T, Schroeder S, Claussen CD. Multidetector-row CT for the noninvasive detection of highgrade coronary artery stenoses and occlusions: first results. Radiology 1999; 213 (P) : 435. [12] Flohr T, Ohnesorge B, Kopp AF, Becker C, Haliburton SS, Knez A. A reconstruction concept for ECG-gated multi-slice spiral CT of the heart with pulse-rate adaptive optimization of spatial and temporal resolution. Radiology 2000; 217 (P): 438. [13] American Heart Association Committee Report. A reporting system on patients evaluated for coronary artery disease. Circulation 1999; 51: 7 34. [14] Schmermund A, Rensing BJ, Sheedy PF, Bell MR, Rumberger JA. Intravenous electron-beam computed tomographic coronary angiography for segmental analysis of coronary artery stenoses. J Am Coll Cardiol 1998; 31: 1547 54. [15] Niemann K, Oudkerk M, Rensing BJ et al. Coronary angiography with multi-slice computed tomography. Lancet 2001; 357: 599 603. [16] Achenbach S, Giesler T, Ropers D et al. Detection of coronary artery stenoses by contrast-enhanced, retrospectively electrocardiographically-gated, multislice spiral computed tomography. Circulation 2001; 103: 2535 8. [17] Ohnesorge B, Flohr T, Becker C, Kopp AF, Knez A, Reiser M. Dose evaluation and dose reduction strategies for ECGgated multi-slice spiral CT of the heart. Radiology 2000; 217 (P): 487.