| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Cardiac Imaging |
1 From the Department of Diagnostic Radiology, University Hospital of Ulm, Steinhoevelstrasse 9, D 89070 Ulm, Germany (M.H.K.H., H.S., F.T.S., H.J.B., A.J.A.); Philips Medical Systems, Cleveland, Ohio (L.D.V.); and Philips Research Laboratories, Hamburg, Germany (R.M., M.G.). Received September 1, 2003; revision requested November 11; final revision received April 7, 2004; accepted April 28. Address correspondence to M.H.K.H. (e-mail: martin.hoffmann@medizin.uni-ulm.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
MATERIALS AND METHODS: The local ethics committee approved the study, and informed consent was obtained from all patients. Fifty patients underwent coronary CT angiography (heart rate range, 45–103 beats per minute). Raw data from helical CT and electrocardiography (ECG) were saved in a combined data set. Retrospectively ECG-gated images were reconstructed at preselected phases (50% and 80%) of the cardiac cycle. A 3D voxel-based approach with cardiac phase weighting was used for reconstruction. Testing for correlation between heart rate, cardiac phase reconstruction window, and image quality was performed with Kruskal-Wallis analysis. Image quality (freedom from cardiac motion–related artifacts) was referenced against findings at conventional angiography in a secondary evaluation step. Regression analysis was performed to calculate heart rate thresholds for future ß-blocker application.
RESULTS: A significant negative correlation was observed between heart rate and image quality (r = 0.80, P < .001). Motion artifact–free images were available for 44 (88%) patients and were achieved consistently at a heart rate of 80 or fewer beats per minute (n = 39). Best image quality was achieved at 75 or fewer beats per minute. Segmental analysis revealed that 97% of arterial segments (diameter 
1.5 mm according to conventional angiography) were assessable at 80 or fewer beats per minute. Premature ventricular contractions and rate-contained arrhythmia did not impede diagnostic assessment of the coronary arteries in 10 (83%) of the 12 patients affected.
CONCLUSION: Motion-free coronary angiograms can be obtained consistently with 16–detector row CT scanners and adaptive multicyclic reconstruction algorithms in patients with heart rates of less than 80 beats per minute.
© RSNA, 2004
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Background
A crucial step for performing coronary CT angiography is to relate the minimum temporal resolution available with the scanner configuration to the cardiac cycle phases in which coronary motion is minimal. Cardiac phase preselection in this study was influenced by the results obtained by Achenbach et al (10) with regard to coronary artery motion on transverse electron-beam CT images. Achenbach et al found that minimal motion occurred in the end-systolic phase, at 48% of the cardiac cycle, and in the middiastolic phase, at 84% of the cardiac cycle. We therefore preselected phases at 50% and 80% of the cardiac cycle for this study. The duration of these motion-free intervals is inversely related to heart rate: The middiastolic phase interval (at approximately 80% of the cardiac cycle) decreases with a curvilinear function against increasing heart rate (11), whereas the end-systolic phase interval (at approximately 50% of the cardiac cycle) remains relatively constant and decreases linearly with increasing heart rate (Fig 1) (12).
According to the physiology of the cardiac cycle, this end-systolic phase (at approximately 50% of the cycle) comprises both reduced ejection during systole and isovolumic relaxation during early diastole. The middiastolic phase (at approximately 80% of the cycle) corresponds to diastasis, the quiescent phase of diastole, which occurs between rapid filling and atrial contraction. If both phases could be reliably detected, reconstruction could be performed by using data from only these two phases, with a total number of 600 images per patient (scan length, 12 cm; z-axis resolution, 0.8 mm; increment, 0.4 mm). This result would represent an improvement over total coverage of the cardiac cycle, which might involve reconstruction of as many as 10 phases and result in 3000 images per patient, necessitating reader-guided empiric phase evaluation.
To identify a specific cardiac phase, a dynamic model or delay algorithm was used (13) that enables modeling of a compliance function according to both a linearly declining constant phase delay (systolic changes) and an exponentially declining relative phase delay (diastolic changes). The model was applied in this study by using so-called physiologic percentages to track cardiac phases (13). Physiologic percentages include a delay offset (systolic linear changes) and a percentage of delay (diastolic exponential changes). This procedure theoretically results in better tracking of phases during scanning, as the heart rate changes during the breath hold. In addition, the delay algorithm should enable better interindividual characterization of heart phases, although all studies of coronary motion during the cardiac cycle have shown large variations in phase duration among patients (10,13,14).
Data Acquisition
Contrast-enhanced multi–detector row CT examinations were performed in the supine position, during a single breath hold, by using a 16–detector row scanner (MX 8000 IDT; Philips Medical Systems). All CT angiographic studies were preceded by a scout acquisition. For bolus tracking, transverse sections were acquired at the level of the aortic root (section position was related to the location of the carina in the scout images). Fifteen transverse sections were acquired successively, one section every 1.7 seconds (120 kV, 30 mAs, 6-mm section thickness). Scanning began with a fixed delay of 10 seconds after contrast material administration. A total of 10 mL of an iodinated contrast agent containing 370 mg of iodine per milliliter (iomeprol, Solutrast 370; Altana, Konstanz, Germany) was given at a rate of 4 mL/sec, followed by a chaser bolus of 30 mL saline solution, given at 3.5 mL/sec. A region of interest (mean diameter, 11.2 mm ± 5.1 [standard deviation]) was positioned in the aortic root, and attenuation enhancement values (in Hounsfield units) within this region were plotted against time (F.T.S., M.H.K.H.).
Coronary CT angiography was timed to coincide with peak contrast enhancement, which was derived from attenuation measurements during bolus tracking. Electrocardiography (ECG) was performed during continuous CT data acquisition. The scanning protocol included a collimation of 16 sections with an individual section thickness of 0.75 mm (16 x 0.75 mm), table feed increment of 6.86 mm/sec, pitch of 0.24, and gantry rotation time of 420 msec). Tube voltage of 140 kV and current of 230 mA (400 mAs) were typically applied. Depending on cardiac dimensions, the scanning time varied between 16 and 23 seconds.
Image Reconstruction
After acquisition of the raw helical CT data, retrospectively ECG-synchronized sections were reconstructed. A multicyclic reconstruction algorithm with capabilities for slight improvement in temporal resolution was used (9). The word multicyclic in this context is synonymous with multisegmental, another term used in the literature; both terms refer to the segmentation of the gating window for data acquisition over multiple heartbeats. This method of segmentation theoretically improves temporal resolution commensurately with increase in the absolute number of cardiac cycles used in image reconstruction. The algorithm used in this study differed from those used in previous studies in that it was integrated with 3D voxel-based backward projection of the cone beam (9,15,16). This difference, we hypothesized, would enable improvements in image quality beyond those possible with other multisegmental approaches of which we were aware, that were investigated prior to this study (15). A description of the reconstruction algorithm is given in the Appendix.
Low-pitch helical CT is performed to image the heart with z-axis redundancy in coverage over multiple cardiac cycles. The range of angles used in acquisition of a voxel, from the first to the last acquisition angle, defines the illumination window (Fig 2). This window determines the amount of data that can be used for the reconstruction of the voxel. The algorithm adaptively uses the maximum number of cardiac cycles available in the illumination window for the reconstruction of that voxel (16).
|
The order of segmentation or number of cardiac cycles used in reconstruction of a certain subvolume depends on the pitch setting in combination with the heart rate (15). With the heart rate remaining constant, the pitch or the table feed will determine how many cardiac cycles are available for an illumination window of a given width. With pitch remaining constant, more cardiac cycles become available with increasing heart rate.
There are two periodically moving objects involved in the imaging process: the gantry, which circles the patient at a maximum speed of one rotation every 0.42 second, and the heart, which beats at a variable rate. The motion of these two objects may be correlated and synchronous, in accordance with so-called frequency harmonics, or may be decorrelated and asynchronous. Segmentation over multiple cardiac cycles is of maximum benefit when motion is fully asynchronous, but it provides little advantage for temporal resolution when motion occurs in harmonic cycles. With multicyclic segmentation, therefore, the curve for mean temporal resolution plotted against increasing heart rate has a wavelike appearance (Fig 1). Depending on the heart rate, temporal resolution varies between 95 and 210 msec (Fig 1).
Because data were acquired continuously and retrospective gating was applied, reconstruction could be carried out by using data from any point within the cardiac cycle. Routinely, two data sets were reconstructed, with gating windows centered at 80% and 50% of the R-R cycle. These positions are referred to in this article as middiastolic (approximately 80% of the cycle) and end-systolic (approximately 50% of the cycle) reconstruction windows. Adjacent positions (± 5%) were used if no satisfactory results were achieved with these initial window positions.
The transverse sections and volume-rendered images were compared side by side at a dedicated workstation for CT image reconstruction and display (MxView, version 4.1; Philips Medical Systems, Cleveland, Ohio), and the image data set with the fewest motion artifacts was selected for further analysis.
If satisfactory results could not be obtained for all coronary artery segments in the same cardiac phase, multiple phases were used for the analysis of the different branches or segments.
Postprocessing and Analysis
The CT data sets were postprocessed at the workstation described earlier. Image analysis was performed by using multiplanar reformation (MPR), maximum intensity projection, volume rendering, and curved MPR techniques, in addition to transverse source image data. Two investigators (M.H.K.H., with 3 years of experience, and H.S., with 1 year of experience, in coronary CT angiography) assessed each coronary artery segment according to guidelines of the American College of Cardiology and the American Heart Association (17), and a decision was achieved by means of direct consensus reading. A first evaluation was performed to assess the possibility of obtaining motion-free images. Transverse image stacks were controlled for continuous depiction (successively matched contours) of the coronary arteries by scrolling through the transverse image data sets at a viewing workstation. The assessment was repeated by using thin-slab maximum intensity projections (mean slab thickness, 3 mm; range, 1–5 mm) in coronal and sagittal imaging planes. Subsequently, curved MPR images were assessed for motion-free imaging (ie, no tattered or ragged delineation) of the coronary branches. Images were generated as follows: Volume-rendered images, with the volume of interest defined to exclude the rib cage and pulmonary structures, were oriented along the z-axis. A line was electronically drawn in the center of the lumen of all major coronary artery branches to generate curved MPR images. At least two curved MPR images were generated for each center line by using perpendicular viewing angles and planes. All three image postprocessing modalities (transverse image stacks, thin-slab maximum intensity projections, and curved MPRs) were used for subjective evaluation. Images were graded for quality of depiction of each vessel, with a five-point scale from 1 (best image quality) to 5 (worst image quality). The scores were defined as shown in Table 1. Curved MPR images and volume-rendered images with various quality scores are shown in Figures 3 and 4.
|
|
|
|
|
|
|
|
|
ECG Data
ECG tracings acquired during scanning were digitized and stored with the unprocessed CT data set. Each ECG tracing, from the start of scanning to the end, was retrospectively assessed for arrhythmic or premature beats (M.H.K.H.). Arrhythmic or premature beats were correlated with the corresponding z-axis imaging plane for assessment of motion artifact induction. An R-wave peak trigger was generated according to a voltage threshold that was automatically adjusted on the basis of the patient’s ECG data. The distance in milliseconds between two R-wave peaks was calculated and plotted against scanning time. Mean R-R interval and standard deviation were calculated for the total scanning duration. A ß-blocker was not applied in any of the examinations. (Patients already receiving a ß-blocker [n = 19] continued with their established regimen.)
Conventional Angiographic Data
Images obtained at conventional angiography were postprocessed at the workstation by using software (eFilm, version 1.8.3; Merge eFilm, Milwaukee, Wis). Image display was calibrated on the basis of the dimensions of the catheter tip (4–6 F) in the coronary artery ostium. Angiograms were evaluated in consensus by a radiologist (M.H.K.H., with 8 years of experience in coronary angiogram reading) and a consulting physician from the cardiology service. All segments of the coronary artery tree with a reference diameter of 1.5 mm or greater were included in the study. Reference diameter was measured in the midsection of each segment, or, in the presence of stenosis in the midsection of the segment, proximal to the stenosis.
Comparisons of CT and conventional angiographic data were performed separately for the subgroup of images that were motion artifact free according to primary subjective analysis. Images in this subgroup were evaluated for completeness of coverage of distal segments and side branches with a reference diameter of at least 1.5 mm. This was done by quantifying the segment diameter on the conventional angiogram and by verifying that all segments with a diameter of 1.5 mm or greater were contrast enhanced on the CT angiogram.
In the subgroup of images with quality scores of 4 or 5 (ie, those on which motion artifacts were seen), all arterial segments for which depiction was affected by motion artifacts were reviewed and evaluated for the presence of stenotic lesions or occlusions that may mimic artifacts. Patients with no available conventional angiographic data (n = 8) were not included in the secondary evaluation. To avoid verification bias, we did not include the results of stenosis evaluation in the study analysis.
Statistical Analysis
Continuous variables were expressed as mean ± standard deviation. A P value of less than .05 indicated a statistically significant difference. Descriptive statistics were stratified according to image quality score in five groups. The Fisher exact test was used to evaluate categorical data (distribution of the sexes among image quality groups). The Kruskal-Wallis test was used to evaluate quantitative parameters (age, heart rate) stratified across the five image quality scores. The Wilcoxon two-sample test was used to compare the selection of phases according to heart rate. Pearson correlation analysis was used to compare the average image quality score for all segments with the heart rate. A linear regression equation was used to calculate heart rate threshold values for future ß-blocker application. Retrospective power analysis was performed to test the correlation coefficient obtained against no correlation (r = 0) and low correlation (r = 0.30) presumed for the general population. Power analysis was conducted for the sample size of 50 patients. All statistical analyses were performed with software (SAS, version 8.02; SAS Institute, Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
CT Image Reconstruction
The two cardiac phases (50% and 80% of the R-R interval) initially used for reconstruction of image data for all patients were sufficient to generate motion-free images for 40 (80%) of 50 patients. Additional phases had to be reconstructed for 10 (20%) patients. For one patient, an additional middiastolic reconstruction was performed at 75% of the R-R interval to achieve best image quality (mean heart rate, 67.3 beats per minute). For nine patients, additional reconstructions were performed in the end-systolic phase, at 45%–60% of the R-R interval, which resulted in motion-free delineation of the coronary arteries in three patients but not in the other six patients. The mean heart rate in patients for whom additional reconstruction phases were required was 76.4 beats per minute ± 5.4. All end-systolic phase frames were summarized at 50% of the R-R interval, and all middiastolic frames, at 80%. A plot of heart rates against phase selection (Fig 5) showed that the middiastolic phase (80% of the R-R interval) yielded the best results at lower heart rates (mean, 63.2 beats per minute ± 8.2) and the end-systolic phase (50% of the R-R interval) yielded the best results at higher heart rates (mean, 82.4 beats per minute ± 8.8; P < .05).
|
|
|
Table 2 shows patient demographic data and lists the segments that were scored as assessable at subjective image quality evaluation and the results of secondary evaluation of CT image data against conventional angiographic data. Overall, 581 (88%) of 658 artery segments were assessable. There was no significant variation in assessability over the three major vessel territories: the right coronary artery (167 [86.5%] of 193), left main and left anterior descending artery (199 [86.5%] of 230), and left circumflex artery (138 [87.3%] of 158) (P > .5). The cause of nonassessability was cardiac motion–related artifact in 55 (8.4%) of 658 coronary artery segments. Blooming artifacts due to calcifications, and streak artifacts induced by coronary stents or intracardiac pacing leads, caused nonassessability in 18 (2.7%) and four (0.6%) of 658 segments, respectively. Artifacts that interfered with coronary artery segment assessment in one patient were caused by intracardiac pacing leads. In all other postsurgical patients (n = 6), artifacts either were not present or did not overlie coronary artery segments.
The two readers who performed image quality assessment (M.H.K.H., H.S.) achieved immediate agreement in 36 (62%) of the 50 cases. The consensus discussion about the remaining 14 (28%) was primarily concerned with the discrimination of image quality scores 1 and 2.
To establish heart rate thresholds for noninvasive coronary angiography, a regression analysis was performed (Fig 7). According to the linear regression equation, motion-free images should be possible in patients with a maximum heart rate of 100 beats per minute. Application of the linear equation to images with a quality score of 2 yields a heart rate threshold of 74.2 beats per minute. In most patients with heart rates at or below this level, good to excellent image quality is achievable by using the protocol used in this study.
|
Arrhythmia
In the study group, nine patients had premature ventricular beats during the image acquisition. No apparent motion artifacts could be related to these arrhythmic beats in eight (89%) of these patients. One patient with atrial fibrillation and a heart rate within the range of 60 to 90 beats per minute was included in the study. His overall image quality score was 2. Only slight banding artifacts were discernible (Fig 8). These did not compromise the diagnostic assessment of the coronary tree. One patient developed a bigeminal heart rhythm pattern during scanning. This patient’s image quality score was 5; motion artifacts did not allow assessment of the coronary arteries (Fig 4c).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Good image quality is achievable at low heart rates by using a mid- to end-diastolic phase frame and shifting to end-systolic frames as heart rate increases. Figure 1 shows the temporal resolution achieved by using a 3D multicyclic reconstruction algorithm (9,16) versus diastolic and systolic reconstruction windows. According to theoretical considerations, both the middiastolic (80% of R-R interval) and the end-systolic (50% of R-R interval) reconstruction windows should coincide with low-motion periods in the coronary artery tree to allow motion-free imaging. The plot shown in Figure 1 was constructed according to the approximation of phase duration by Herzog et al (23). The point at which a shift from the diastolic frame to the systolic frame becomes necessary occurs at approximately 65 beats per minute. The results of phase selection in our patient group, however, suggest that the shift from the middiastolic phase (at low heart rates) to the end-systolic phase (at higher heart rates) should take place at 72–80 beats per minute. To explain this discrepancy, we can only postulate that the diastolic phase duration according to the theoretical considerations of Herzog et al was underestimated (23).
Besides motion artifacts, other factors, such as the presence of high-attenuation structures adjacent to the coronary arteries, may impede accurate lumen assessment. Calcified plaque and extensive calcification of coronary artery walls appear oversized on CT images (24). This so-called blooming artifact is supposedly related to partial-volume averaging effects caused by insufficient spatial resolution. With 16–detector row CT, extensive calcifications prevented segment assessment to a lesser extent than with four–detector row CT (in 2.7% vs 5.0% of segments) (8). The validity of this comparison, however, is limited by the lack of available data with regard to calcium prevalence stratification in the compared groups. Nevertheless, the comparison of these results indicates that the slight improvement in spatial resolution achievable with the 16–detector row scanner may reduce blooming artifacts to some extent. Compared with voxel size for four–detector row scanners, which is slightly greater than 1 mm in all dimensions, effective voxel size for 16–detector row scanners is reduced to 0.6 x 0.6 x 0.8 mm.
High-attenuation objects inside the cardiac cavities (eg, pacing lead tips) or adjacent to the cardiac structures (eg, sternal wires after surgery or metallic clips adjacent to mammary pedicles after coronary artery bypass grafting) are a further potential source of image artifacts. These artifacts, however, were sufficiently mitigated with cone-beam correction included in the reconstruction algorithm (9).
Comparison with Other Studies
A regression plot similar to that shown in Figure 7, relating qualitative image scores to heart rate, is available in a published report of a four–detector row scanner study. Hong et al (18) calculated an upper heart rate limit of 74.5 beats per minute for achievement of motion-free images with the four–detector row scanner. If the same method of calculation used by Hong et al were applied to the data set in this study (ie, if the worst image score associated with motion-free image quality were entered into the linear regression equation), we would come up with a threshold of 100 beats per minute. However, a motion-free image at this quality level would not offer the best diagnostic value, as it would include either compromised distal coverage or blurred lumen definition because of higher noise levels. To obtain optimal diagnostic results, images must have a quality score of 2 or higher. If this image quality level is factored into the regression equation, then we find that optimal diagnostic results are achievable at heart rates of up to 75 beats per minute. In other words, regarding the maximization of diagnostic image quality, the results with 16–detector row scanners do not seem to differ from those with four–detector row scanners. As we do not have access to the image database of Hong et al, however, we cannot determine whether overall image quality in our group was improved.
Other studies performed with four–detector row CT scanners indicate that patients must have a heart rate of no more than 65 beats per minute to obtain motion-free images (7). This was the threshold extensively accepted by others (1,3,20,25).
The approach recommended by Kopp et al (26), the reconstruction of multiple phases for every patient, does not seem necessary with the protocol used in this study. Although reconstruction time has been tremendously reduced, the amount of data generated may be difficult to handle, both for the radiologist and for the computerized network server or storage hardware.
The data obtained in this study suggest that motion-free imaging is reliably possible in patients with a heart rate of less than 80 beats per minute; the results of regression analysis and theoretical considerations suggest much higher thresholds. One factor that might explain this discrepancy between our image data and the results of our statistical analysis is table feed. In this study, we used a fixed table feed of 6.86 mm/sec, whereas the approach currently used to optimize temporal resolution is to vary the table feed settings. The adaptive multicyclic reconstruction algorithm is not discrete; it uses as many data as possible, depending on factors such as heart rate and table feed (9). With heart rate–adapted table feed settings, we might therefore in the future be able to achieve motion-free imaging at higher heart rates, as indicated by the results of regression analysis.
Limitations
The approach we used in reconstructing image data of the coronary arteries was an adaptation of the simple approach suggested by Achenbach et al (10), who investigated the in-plane motion of the coronary arteries by using high-resolution electron-beam CT (50-msec temporal resolution); the pattern of motion in the coronary arteries, however, is truly 3D (13,14). True 3D motion was not accounted for in the phase selection method used in this study. The simple two-dimensional approach of Achenbach et al proved sufficient in 80% of cases, and phase selection had to be empirically modified only in the remaining 20%.
We were not able to apply an absolute or relative phase location algorithm for comparison with the physiologic phase selection algorithm. The overall performance in patients with rate-contained arrhythmia (ie, atrial fibrillation not exceeding a heart rate of 80–90 beats per minute in any single cardiac cycle), and the good results with regard to sporadic premature heartbeats, indicate a potential advantage with use of the physiologic phase-delay algorithm. Additional studies are warranted to assess this effect more accurately.
Another limitation of the current study was the incomplete conventional angiographic data set. Eight patients who were scheduled to undergo conventional coronary angiography after multi–detector row CT refused the invasive procedure because of negative findings at CT. They were informed that the literature shows a 97% negative predictive value for CT angiographic findings (3,22). They subsequently decided to undertake no further diagnostic studies. Stenosis detection assessment based on this incomplete data set could be biased and was therefore omitted from the study design.
Implications for Coronary Angiographic Imaging
The temporal resolution of images obtained with 16–detector row CT and reconstruction based on a single cardiac cycle is only slightly improved compared with that of images obtained with four–detector row CT, with improvement attributable solely to an increase in gantry rotation speed from 0.50 second to 0.42 second. Further improvements gained with multicyclic reconstruction are translated into substantial increases in resolution only at certain heart rates (9). Because of highly variable heart rates even during the short breath-hold duration applied in the current study, any strategy to optimize temporal resolution on the basis of a single fixed heart rate will fail. Variability in temporal resolution due to inconsistent heart rate during the breath hold may explain the only slight total increase in the maximum heart rate threshold for motion-free imaging with 16–detector row scanners compared with four–detector row scanners. Relative to a normal resting heart rate range of 50–90 beats per minute (27), however, a maximum heart rate threshold of 80 beats per minute will allow motion-free imaging without ß-blocker application in the majority of patients with a normal sinus rhythm.
The preselection of heart phases as suggested by the data of Achenbach et al proved a very effective means of optimizing the postprocessing strategy. The use of the dynamic delay model and so-called physiologic phases enabled the accommodation of premature heartbeats and rate-contained arrhythmia without user interference. Whether this method offers advantages over currently applied simple absolute or relative phase-delay algorithms remains to be shown in future studies.
The data from this study suggest that ß-blocker–induced reduction of the resting heart rate to 65 or fewer beats per minute may not be necessary for coronary CT angiography performed with 16–detector row scanners and multicyclic reconstruction, which provide consistent motion-free image quality at heart rates up to 80 beats per minute. Because the heart rate increases slightly during scanning in most patients, however, it is suggested that the initial heart rate be reduced to less than 75 beats per minute to include a safety margin and to optimize image quality.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
As a first step, the helical projection data are re-sorted to create a virtual parallel-beam geometry (see Figure A1). This process is referred to as parallel rebinning (15). Afterward, the projection data are weighted with the cosine of the cone-angle and filtered with a one-dimensional band-limited high-pass filter (15). Finally, the preprocessed data are subjected to 3D backward projection along the true paths of the rays. During the final backward-projection step, the data are weighted according to a normalized weighting function. This weighting function, wall(
, 
), which depends on each voxel and on the transverse projection angle , enables the retrospectively gated image reconstruction according to data such as those from ECG.
|
, ). The smooth weighting function is used to reduce motion artifacts and to minimize discontinuities between the normalized weighting profiles of the voxels.
To perform retrospective gating of the projection data, a second weighting function, wc(), is introduced, called the cardiac weighting function. Data around the selected cardiac phase are weighted with positive values, whereas data that do not correspond to the selected cardiac phase are discarded with zero weighting. The width of the cardiac weighting function, or in other words the width of the cardiac gating window, defines the data set used for the backward projection and hence primarily defines the temporal resolution. To obtain the maximum temporal resolution, the widths of the cardiac weighting functions are optimized for each cardiac cycle separately. This is achieved by minimizing the cardiac gating window to the smallest width that still ensures sufficient projection data for reconstruction.
The final weighting function wall(, ) is derived from the combination of the illumination weighting function wil(, ) and cardiac weighting function wc() by using a normalizing approach:
|
N (for a more detailed description, see reference 16). The use of a low-pitch scanning mode guarantees the acquisition of a reasonable amount of redundant data, which allows the gated image reconstruction. With the normalization approach, the algorithm automatically adapts the number of cardiac cycles used for the reconstruction of every single voxel, enabling reconstruction of data from individual or multiple cardiac cycles. The temporal resolution depends on the weighted contribution of the projection data used to reconstruct a voxel, the so-called sensitivity profile (16). The temporal resolution of a volume covering the heart is calculated as the mean temporal resolution for all voxels within the volume.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, H.J.B., L.D.V., A.J.A., M.H.K.H.; study concepts, M.H.K.H., M.G., L.D.V.; study design, M.H.K.H., R.M.; literature research, M.H.K.H., H.S., F.T.S.; clinical studies, M.H.K.H., F.T.S.; data acquisition and analysis/interpretation, H.S., F.T.S., M.H.K.H.; statistical analysis, M.H.K.H., M.G.; manuscript preparation, M.H.K.H., R.M.; manuscript definition of intellectual content, A.J.A.; manuscript editing, F.T.S., R.M.; manuscript revision/review, A.J.A., M.G., R.M.; manuscript final version approval, M.H.K.H., M.G., A.J.A., H.J.B., L.D.V.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Baumuller, S. Leschka, L. Desbiolles, P. Stolzmann, H. Scheffel, B. Seifert, B. Marincek, and H. Alkadhi Dual-Source versus 64-Section CT Coronary Angiography at Lower Heart Rates: Comparison of Accuracy and Radiation Dose Radiology, July 8, 2009; (2009) 2531090065. [Abstract] [Full Text]
|
|
|
|
 |
|
 R. Donnino, J. E. Jacobs, J. V. Doshi, E. M. Hecht, D. C. Kim, J. S. Babb, and M. B. Srichai Dual-Source Versus Single-Source Cardiac CT Angiography: Comparison of Diagnostic Image Quality Am. J. Roentgenol., April 1, 2009; 192(4): 1051 - 1056. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L ROMANO, L GRAZIOLI, L BONOMO, J-R XU, K-M CHEN, R DORE, A VANZULLI, and C CATALANO Enhancement and safety of iomeprol-400 and iodixanol-320 in patients undergoing abdominal multidetector CT Br. J. Radiol., March 1, 2009; 82(975): 204 - 211. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. A. Herzog, L. Husmann, N. Burkhard, O. Gaemperli, I. Valenta, F. Tatsugami, C. A. Wyss, U. Landmesser, and P. A. Kaufmann Accuracy of low-dose computed tomography coronary angiography using prospective electrocardiogram-triggering: first clinical experience Eur. Heart J., December 2, 2008; 29(24): 3037 - 3042. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Matsumoto, T. Kondo, S. Watanabe, R. Kikumoto, T. Shimada, Y. Hiraoka, and K. Ueda ECG-Edited Middiastolic Phase Reconstruction Improves Image Quality at 64-MDCT Coronary Angiography of Patients with Atrial Fibrillation Am. J. Roentgenol., December 1, 2008; 191(6): 1659 - 1666. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Bastarrika, C. N. De Cecco, M. Arraiza, S. Mastrobuoni, J. C. Pueyo, M. Ubilla, and G. Rabago Dual-Source CT for Visualization of the Coronary Arteries in Heart Transplant Patients with High Heart Rates Am. J. Roentgenol., August 1, 2008; 191(2): 448 - 454. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. A. Bluemke, S. Achenbach, M. Budoff, T. C. Gerber, B. Gersh, L. D. Hillis, W. G. Hundley, W. J. Manning, B. F. Printz, M. Stuber, et al. Noninvasive Coronary Artery Imaging: Magnetic Resonance Angiography and Multidetector Computed Tomography Angiography: A Scientific Statement From the American Heart Association Committee on Cardiovascular Imaging and Intervention of the Council on Cardiovascular Radiology and Intervention, and the Councils on Clinical Cardiology and Cardiovascular Disease in the Young Circulation, July 29, 2008; 118(5): 586 - 606. [Full Text] [PDF]
|
|
|
|
|
|
N. Matsuura, J. Horiguchi, H. Yamamoto, N. Hirai, T. Tonda, N. Kohno, and K. Ito Optimal Cardiac Phase for Coronary Artery Calcium Scoring on Single-Source 64-MDCT Scanner: Least Interscan Variability and Least Motion Artifacts Am. J. Roentgenol., June 1, 2008; 190(6): 1561 - 1568. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Leschka, H. Scheffel, L. Husmann, O. Gamperli, B. Marincek, P. A. Kaufmann, and H. Alkadhi Effect of Decrease in Heart Rate Variability on the Diagnostic Accuracy of 64-MDCT Coronary Angiography Am. J. Roentgenol., June 1, 2008; 190(6): 1583 - 1590. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Brodoefel, C. Burgstahler, I. Tsiflikas, A. Reimann, S. Schroeder, C. D. Claussen, M. Heuschmid, and A. F. Kopp Dual-Source CT: Effect of Heart Rate, Heart Rate Variability, and Calcification on Image Quality and Diagnostic Accuracy Radiology, May 1, 2008; 247(2): 346 - 355. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Aziz, K. Berger, K. Claycombe, R. Huang, R. Patel, and G. S. Abela Noninvasive Detection and Localization of Vulnerable Plaque and Arterial Thrombosis With Computed Tomography Angiography/Positron Emission Tomography Circulation, April 22, 2008; 117(16): 2061 - 2070. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Achenbach, U. Ropers, A. Kuettner, K. Anders, T. Pflederer, S. Komatsu, W. Bautz, W. G. Daniel, and D. Ropers Randomized comparison of 64-slice single- and dual-source computed tomography coronary angiography for the detection of coronary artery disease. J. Am. Coll. Cardiol. Img., March 1, 2008; 1(2): 177 - 186. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Prat-Gonzalez, J. Sanz, and M. J. Garcia Cardiac CT: Indications and Limitations J. Nucl. Med. Technol., March 1, 2008; 36(1): 18 - 24. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. T. Johnson, J. Eng, H. K. Pannu, and E. K. Fishman 64-MDCT Angiography of the Coronary Arteries: Nationwide Survey of Patient Preparation Practice Am. J. Roentgenol., March 1, 2008; 190(3): 743 - 747. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Oncel, G. Oncel, and A. Tastan Effectiveness of Dual-Source CT Coronary Angiography for the Evaluation of Coronary Artery Disease in Patients with Atrial Fibrillation: Initial Experience Radiology, December 1, 2007; 245(3): 703 - 711. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Seifarth, S. Wienbeck, M. Pusken, K.-U. Juergens, D. Maintz, C. Vahlhaus, W. Heindel, and R. Fischbach Optimal Systolic and Diastolic Reconstruction Windows for Coronary CT Angiography Using Dual-Source CT Am. J. Roentgenol., December 1, 2007; 189(6): 1317 - 1323. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Husmann, S. Leschka, L. Desbiolles, T. Schepis, O. Gaemperli, B. Seifert, P. Cattin, T. Frauenfelder, T. G. Flohr, B. Marincek, et al. Coronary Artery Motion and Cardiac Phases: Dependency on Heart Rate Implications for CT Image Reconstruction Radiology, November 1, 2007; 245(2): 567 - 576. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Dewey, H.-P. Dubel, T. Schink, G. Baumann, and B. Hamm Head-to-head comparison of multislice computed tomography and exercise electrocardiography for diagnosis of coronary artery disease Eur. Heart J., October 2, 2007; 28(20): 2485 - 2490. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Matt, H. Scheffel, S. Leschka, T. G. Flohr, B. Marincek, P. A. Kaufmann, and H. Alkadhi Dual-Source CT Coronary Angiography: Image Quality, Mean Heart Rate, and Heart Rate Variability Am. J. Roentgenol., September 1, 2007; 189(3): 567 - 573. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Rispler, Z. Keidar, E. Ghersin, A. Roguin, A. Soil, R. Dragu, D. Litmanovich, A. Frenkel, D. Aronson, A. Engel, et al. Integrated Single-Photon Emission Computed Tomography and Computed Tomography Coronary Angiography for the Assessment of Hemodynamically Significant Coronary Artery Lesions J. Am. Coll. Cardiol., March 13, 2007; 49(10): 1059 - 1067. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Ghostine, C. Caussin, B. Daoud, M. Habis, E. Perrier, D. Pesenti-Rossi, A. Sigal-Cinqualbre, C.-Y. Angel, B. Lancelin, A. Capderou, et al. Non-Invasive Detection of Coronary Artery Disease in Patients With Left Bundle Branch Block Using 64-Slice Computed Tomography J. Am. Coll. Cardiol., November 21, 2006; 48(10): 1929 - 1934. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Achenbach Computed Tomography Coronary Angiography J. Am. Coll. Cardiol., November 21, 2006; 48(10): 1919 - 1928. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Leschka, S. Wildermuth, T. Boehm, L. Desbiolles, L. Husmann, A. Plass, P. Koepfli, T. Schepis, B. Marincek, P. A. Kaufmann, et al. Noninvasive Coronary Angiography with 64-Section CT: Effect of Average Heart Rate and Heart Rate Variability on Image Quality Radiology, November 1, 2006; 241(2): 378 - 385. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Budoff, S. Achenbach, R. S. Blumenthal, J. J. Carr, J. G. Goldin, P. Greenland, A. D. Guerci, J. A.C. Lima, D. J. Rader, G. D. Rubin, et al. Assessment of Coronary Artery Disease by Cardiac Computed Tomography: A Scientific Statement From the American Heart Association Committee on Cardiovascular Imaging and Intervention, Council on Cardiovascular Radiology and Intervention, and Committee on Cardiac Imaging, Council on Clinical Cardiology Circulation, October 17, 2006; 114(16): 1761 - 1791. [Full Text] [PDF]
|
|
|
|
 |
|
 M. Dewey, F. Teige, D. Schnapauff, M. Laule, A. C. Borges, K.-D. Wernecke, T. Schink, G. Baumann, W. Rutsch, P. Rogalla, et al. Noninvasive detection of coronary artery stenoses with multislice computed tomography or magnetic resonance imaging. Ann Intern Med, September 19, 2006; 145(6): 407 - 415. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Sigurdsson, P. Carrascosa, M. H. Yamani, N. L. Greenberg, S. Perrone, G. Lev, M. Y. Desai, and M. J. Garcia Detection of Transplant Coronary Artery Disease Using Multidetector Computed Tomography With Adaptative Multisegment Reconstruction J. Am. Coll. Cardiol., August 15, 2006; 48(4): 772 - 778. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. K. Pannu, J. E. Jacobs, S. Lai, and E. K. Fishman Coronary CT angiography with 64-MDCT: assessment of vessel visibility. Am. J. Roentgenol., July 1, 2006; 187(1): 119 - 126. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. S. White Invited Commentary RadioGraphics, July 1, 2006; 26(4): 979 - 980. [Full Text] [PDF]
|
|
|
|
 |
|
 C Peebles Computed tomographic coronary angiography: how many slices do you need? Heart, May 1, 2006; 92(5): 582 - 584. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Y. Kim, J. B. Seo, K.-H. Do, J.-N. Heo, J. S. Lee, J.-W. Song, Y. H. Choe, T. H. Kim, H. S. Yong, S. I. Choi, et al. Coronary Artery Anomalies: Classification and ECG-gated Multi-Detector Row CT Findings with Angiographic Correlation. RadioGraphics, March 1, 2006; 26(2): 317 - 333. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Achenbach and W. G. Daniel Computed Tomography of the Coronary Arteries: More Than Meets the (Angiographic) Eye J. Am. Coll. Cardiol., July 5, 2005; 46(1): 155 - 157. [Full Text] [PDF]
|
|
|
|
 |
|
 M. H. K. Hoffmann, H. Shi, B. L. Schmitz, F. T. Schmid, M. Lieberknecht, R. Schulze, B. Ludwig, U. Kroschel, N. Jahnke, W. Haerer, et al. Noninvasive Coronary Angiography With Multislice Computed Tomography JAMA, May 25, 2005; 293(20): 2471 - 2478. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
= outlier.
