DOI: 10.1148/radiol.2341040439
(Radiology 2005;234:35-43.)
© RSNA, 2005
Calcified Coronary Artery Plaque Measurement with Cardiac CT in Population-based Studies: Standardized Protocol of Multi-Ethnic Study of Atherosclerosis (MESA) and Coronary Artery Risk Development in Young Adults (CARDIA) Study1
J. Jeffrey Carr, MD, MSCE,
Jennifer Clark Nelson, PhD,
Nathan D. Wong, PhD,
Michael McNitt-Gray, PhD,
Yadon Arad, MD,
David R. Jacobs, Jr, PhD,
Stephan Sidney, MD, MPH,
Diane E. Bild, MD, MPH,
O. Dale Williams, PhD, MPH and
Robert C. Detrano, MD, PhD
1 From the Departments of Radiology and Public Health Sciences, Wake Forest University School of Medicine, Medical Center Blvd, Winston-Salem, NC 27157 (J.J.C.); Department of Biostatistics, University of Washington Center for Health Studies, Group Health Cooperative, Seattle, Wash (J.C.N.); Heart Disease Prevention Program, University of California, Irvine, Calif (N.D.W.); Department of Radiology, School of Medicine, University of California, Los Angeles, Calif (M.M.G.); Department of Clinical Medicine, Columbia University, New York, NY (Y.A.); Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, Minn (D.R.J.); Kaiser Permanente Research Division, Oakland, Calif (S.S.); Division of Epidemiology and Clinical Applications, National Heart, Lung, and Blood Institute, Bethesda, Md (D.E.B.); Division of Preventive Medicine, University of Alabama at Birmingham, Birmingham, Ala (O.D.W.); and Harbor-UCLA Research and Education Institute, Los Angeles, Calif (R.C.D.). Received March 5, 2004; revision requested May 13; revision received June 1; accepted June 18. Supported by contracts N01-HC-95159 through N01-HC-95169, as well as N01-HC-48047 through N01-HC-48050 and N01-HC-95095, with the National Heart, Lung, and Blood Institute. Additional support was provided by General Clinical Research Center, Wake Forest University Health Sciences, grant M01-RR07122. Address correspondence to J.J.C.
 |
ABSTRACT
|
|---|
Calcified coronary artery plaque, measured at cardiac computed tomography (CT), is a predictor of cardiovascular disease and may play an increasing role in cardiovascular disease risk assessment. The Multi-Ethnic Study of Atherosclerosis (MESA) and the Coronary Artery Risk Development in Young Adults (CARDIA) study of the National Heart, Lung, and Blood Institute are population-based studies in which calcified coronary artery plaque was measured with electron-beam and multi–detector row CT and a standardized protocol in 6814 (MESA) and 3044 (CARDIA study) participants. The studies were approved by the appropriate institutional review board from the study site or agency, and written informed consent was obtained from each participant. Participation in the CT examination was high, image quality was good, and agreement for the presence of calcified plaque was high (
= 0.92, MESA; = 0.77, CARDIA study). Extremely high agreement was observed between and within CT image analysts for the presence ( > 0.90, all) and amount (intraclass correlation coefficients, >0.99) of calcified plaque. Measurement of calcified coronary artery plaque with cardiac CT is well accepted by participants and can be implemented with consistently high-quality results with a standardized protocol and trained personnel. If predictive value of calcified coronary artery plaque for cardiovascular events proves sufficient to justify screening a segment of the population, then a standardized cardiac CT protocol is feasible and will provide reproducible results for health care providers and the public.
© RSNA, 2005
|
INTRODUCTION
|
|---|
Subclinical coronary artery atherosclerosis is highly prevalent among adults in industrialized societies and precedes the occurrence of clinical cardiovascular disease. Calcified coronary artery plaques are a component of atherosclerosis that can be quantified by using cardiac computed tomography (CT). The medical and scientific community has requested scientific validation of the importance of coronary artery calcium quantification (1,2). The Multi-Ethnic Study of Atherosclerosis (MESA) and the Coronary Artery Risk Development in Young Adults (CARDIA) study are population-based cohort studies that include the CT measurement of calcified coronary artery plaque as a means of furthering our understanding of the factors that influence cardiovascular disease and its outcomes (3,4). The CARDIAstudy began in 1986 with black and white men and women who were 18–30 years old and who were 33–45 years old at the time of examination (between May 2000 and September 2001). The MESA began in 2000 with Asian (Chinese), black, Hispanic, and white men and women who were 45–84 years old at the time of examination (between July 2000 and July 2002). Standardized data collection protocols were used to obtain cardiac CT scans at six MESA communities (Baltimore, Md; Chicago, Ill; Los Angeles, Calif; Minneapolis, Minn; New York, NY; and Winston-Salem, NC) and at four CARDIA study communities (Birmingham, Ala; Chicago, Ill; Minneapolis, Minn; South San Francisco, Calif). At a central CT reading center at Harbor-UCLA Research and Education Institute, Los Angeles, Calif, all activities were coordinated and measurements of calcified coronary artery plaque were performed for both studies. In this article, we describe in detail the method of measurement used and its application to the cardiac CT examination in these two studies.
|
PROTOCOL DEVELOPMENT
|
|---|
The steering committees for MESA and the CARDIA study established a joint MESA–CARDIA Study Committee composed of experts in cardiac CT at each of the scanning sites, statisticians from the coordinating centers, and epidemiologists from various sites and from the National Heart, Lung, and Blood Institute, Bethesda, Md, project offices for both the CARDIA study and MESA. This committee evaluated the existing literature and in consensus made decisions and recommendations related to the cardiac CT examination. The studies were approved by the appropriate institutional review board from the study site or agency, and written informed consent was obtained from each participant.
|
PARTICIPANT PREPARATION
|
|---|
Designing the CT examination to be efficient and simple for the participant and the CT technologist reduced participant burden, enhanced protocol standardization across scanning sites, and made possible the integration of the participants’ research study scans into busy clinical scanning schedules. The examination involved obtaining two consecutive scans per participant and was designed to require less than 15 minutes of CT room time. For this examination, scanning in participants was performed without administration of oral or intravenous contrast agents. The two sequential cardiac scans were obtained by using a single breath hold at end inspiration. This approach helped to reduce motion artifacts from breathing and improved image quality in the distal coronary artery circulation by depressing the diaphragm and liver, thereby leading to a reduction in beam attenuation. Participants were instructed about breath holding and its importance to the CT examination.
|
CT EQUIPMENT
|
|---|
Electron-beam CT and four–detector row CT were used in both studies. At two of four CARDIA study sites and three of six MESA sites, CT examinations were performed with electron-beam CT (Table 1). At electron-beam CT, x-rays are created with an electron gun to sweep a stream of electrons onto stationary tungsten rings that partially encircle the patient. The x-rays then pass through the patient in a 216° arc for each imaging level and are used to generate a CT image with a partial scan reconstruction algorithm. At four–detector row CT, an x-ray tube that is mounted on a rotating gantry is used to produce x-rays. The x-rays produced by the tube pass through the patient and are reconstructed by using a partial-scan reconstruction algorithm, but with 220° of data. For both technologies, the images are acquired in an axial scan mode, which is alternatively termed sequential or step-and-shoot mode. Prospective electrocardiographic (ECG) triggering, in which the x-ray beam is turned on and off according to the ECG signal, was used at all sites except one, a CARDIA study site where retrospective ECG gating was used with an axial scan mode. In this mode, the x-ray beam is turned on for two full rotations (1.6 seconds) around the patient. Then the data are retrospectively sorted according to the ECG signal recorded during the x-ray exposure to create an image during the desired phase of the cardiac cycle. With either approach, the four–detector row CT scanners acquired four image sections simultaneously. The four–detector row CT gantry rotation period was 0.5 second for all systems except the system that used retrospective gating; the gantry rotation period for that system was 0.8 second. The similarities and differences between electron-beam CT and multi–detector row CT have been detailed previously (5,6).
|
CT TECHNICAL FACTORS
|
|---|
Several CT units were used in the studies. The electron-beam CT system, Imatron C150, operated with an exposure time of 100 msec, a fixed peak voltage of 130 kVp, and a fixed tube current of 630 mA. The Volume Zoom four–detector row CT system was operated in the axial scan mode with prospective ECG triggering, 140 kVp, and gantry rotation speed of 0.5 second. With this scanner, four 2.5-mm sections were acquired per cardiac cycle. Because the x-ray beam was triggered prospectively with the ECG signal, only a partial scan was acquired, and the actual exposure time (the time that the x-ray beam was on per acquisition) was 360 msec. Two generations of the LightSpeed four–detector row CT systems, LightSpeed QXi and LightSpeed Plus, were used. With both, four 2.5-mm sections were acquired simultaneously. The LightSpeed QXi has a minimum gantry rotation period of 0.8 second, whereas the LightSpeed Plus has a minimum gantry rotation period of 0.5 second. Both of these systems were operated in the axial scan mode (cine) with 120 kVp, and four 2.5-mm sections were acquired simultaneously. The LightSpeed Plus system was used with prospective ECG triggering with an exposure time of 330 msec. The LightSpeed QXi system was used with retrospective gating in the axial scan mode with an extended x-ray exposure designed to acquire images throughout the entire cardiac cycle (cine time set to 1.6 seconds). Only a single image with an effective exposure of 520 msec from the desired cardiac phase, however, was analyzed for the amount of calcified plaque.
Image quality is related to the photon flux of x-rays through the anatomy being imaged and ultimately recorded by the CT detectors. The energy and number of x-ray photons are determined by the x-ray beam energy (tube voltage expressed in peak kilovolts) and the product of the tube current (expressed in milliamperes) multiplied by time (expressed in seconds), which is expressed in milliampere-seconds. These factors also determine radiation exposure. For the electron-beam CT systems, these factors are fixed. The four–detector row CT systems allow users to adjust the tube current and thus the flux of x-rays and subsequent patient dose as appropriate for the imaging task. This capability makes it possible to increase the tube current with patient size to maintain image quality. For the MESA–CARDIA study protocol for four–detector row CT, a two-level setting of tube current that was based on body weight was used. Individuals who weighed 100 kg (220 lb) or less underwent CT with the standard tube current setting, and those who weighed more than 100 kg underwent CT with a tube current setting that was 25% higher than the standard setting (Table 1).
The images were reconstructed into a display field of view of 350 mm (35 cm) to include a calibration phantom, which was positioned under the thorax of each participant. Reconstruction algorithms were specified to be standard nonenhanced algorithms (ie, Imatron C150, normal; LightSpeed QXi and LightSpeed Plus, standard; Volume Zoom, B30f). The nominal section thickness was 3.0 mm for electron-beam CT and 2.5 mm for four–detector row CT. Spatial resolution can be described by the smallest volume element, or voxel, for the protocol for each system and was 1.15 mm3 for four–detector row CT (0.68 x 0.68 x 2.50 mm) and 1.38 mm3 for electron-beam CT (0.68 x 0.68 x 3.00 mm).
|
CARDIAC ECG GATING METHOD
|
|---|
Cardiac ECG gating reduces errors of measurement of calcified coronary artery plaque by synchronizing the time of image acquisition for each level of the scan to a specific phase of the cardiac cycle. The instantaneous velocity of the coronary arteries varies throughout the cardiac cycle, and it varies according to vessel and vessel segment and among individuals (7,8). The ECG triggering was set at 80% of the R-R interval for electron-beam CT and at 50% of the R-R interval for four–detector row CT (Fig 1). The four–detector row CT systems acquired four images per cardiac cycle, and the electron-beam CT systems acquired one image per cardiac cycle. In the CARDIA study, the 0.8-second LightSpeed QXi system used retrospective cardiac gating, and the image corresponding to 50% of the R-R interval was selected by using an automated program designed for this purpose (SmartScore; GE Medical Systems) in a postprocessing step.
View larger version (20K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. ECG obtained during a cardiac CT examination illustrates ECG triggering points and temporal resolution of CT systems. In this example, heart rate (60 beats per minute) results in the cardiac cycle (R-R interval) of 1000 msec. For four-detector row CT (MDCT) systems, image acquisition began after 50% of the R-R interval with a temporal window of approximately 250-300 msec for all systems except for those at one scanning site in CARDIA study; the temporal window for systems at that center was approximately 520 msec. For electron-beam CT (EBCT) systems, image acquisition began at 80% of the R-R interval with a temporal window of 100 msec.
|
|
|
RADIATION DOSIMETRY
|
|---|
Radiation exposure to participants is inherent in the technique of CT, in which x-rays are used to generate in vivo images noninvasively. With prospective cardiac ECG gating, irradiation of the patient occurs only during a fraction of the cardiac cycle and produces the lowest exposure. Retrospective gating results in a higher exposure directly proportional to the longer exposure time required. Retrospective gating was performed only with the LightSpeed QXi system, as prospective cardiac gating was not possible with this system at the time of the study. Effective dose provides an estimate of exposure that takes into account the amount of radiation, as well as the sensitivity to radiation of the organs that are exposed. Technically, the effective dose, measured in sieverts, is described as "an estimate of the uniform, whole-body equivalent dose that would produce the same level of risk for adverse effects that results from the non-uniform partial body irradiation" (9). The effective dose allows direct comparison with other sources of radiation exposure, which include natural background (3–3.6 mSv/y) and allowable workplace exposures (50 mSv/y), and is the preferred measure of exposure with CT (9–14). By using these methods, estimates of effective dose according to CT scanner were calculated for a standard male and female subject. The higher estimates in women are attributable to breast tissue in the scanned region. Estimates for men and women, respectively, that were determined by using the MESA–CARDIA study protocol for a single scan obtained through the heart for the Imatron C150, Volume Zoom, LightSpeed Plus, and LightSpeed QXi were as follows: 0.6 and 0.7, 0.9 and 1.1, 1.5 and 1.9, and 4.6 and 5.6 mSv. These values are also presented in Table 1. The consent forms, which had been previously approved by the local investigational review boards, provided the participants with estimates of radiation exposure in lay language.
|
MEASUREMENT OF CALCIFIED CORONARY ARTERY PLAQUE
|
|---|
During the CT examinations at the CT reading center, two consecutive scans (scans 1 and 2) of the heart were obtained; these scans were independently analyzed for calcified coronary artery plaque and were assessed for quality by CT image analysts (or observers). The CT reading center investigators (R.C.D. and N.D.W.) supervised the initial training and quality control of the CT image analysts and the reading process. The image analysts were dedicated research personnel (graduates of non-U.S. medical schools) who were selected on the basis of their prior experience and training and who were trained by the two investigators just indicated in the technical aspects of using the software and measuring calcified plaque. To reduce between–image analyst (observer) variability, only two individuals served as image analysts in each study.
At the CT reading center, custom software was developed to measure calcified plaque specifically for these studies; the custom software was based on software used during the year 10 examination in 1994–1995 of the CARDIA study (15). Key aspects of the software and reading process include the blinding of CT image analysts to any clinical information about study participants and to the calculated results (eg, Agatston score). In the reading process, batches of up to 250 scans on an electronic work list were randomized, thereby reducing the probability that the CT image analysts would sequentially process the two scans of a participant. The software automatically checks the technical parameters (eg, tube voltage expressed in peak kilovolts, tube current expressed in milliamperes, and field of view) used to obtain the scan and notifies the image analyst if values are outside those specified in the protocol. On the CT images, the software automatically locates the four calibration phantom standards and measures the CT attenuation of each calibration phantom by computing the mean CT number for pixels contained within each of the four 15-mm-diameter regions of interest on the image. These data are used to calibrate the image to a standardized level across all study sites.
The image analysts identify the anatomic course of the coronary arteries on the CT images by assigning waypoints along the length of the major arteries. The waypoints are used by the software, along with image data, to define a line corresponding to the trajectory of the coronary artery across the surface of the heart. The program calculates and then displays the three-dimensional course of each coronary artery trajectory in the image data. The image analysts review the coronary artery trajectories determined by the program and adjust the computer-generated trajectory if it deviates from the observed course of the coronary artery. The coronary artery trajectories allow quantification and location of calcified plaque within the coronary arteries and are saved to facilitate future analysis. By using the coronary artery trajectories, the software automatically identifies candidate calcified plaques on the basis of predefined minimum criteria for CT attenuation (130 HU), minimum calcified plaque size (4.6 mm3, four–detector row CT; 5.5 mm3, electron-beam CT), and distance from the coronary artery trajectory (location within an 8-mm radius of the trajectory). The image analyst systematically reviews each candidate calcified plaque and either accepts or rejects its inclusion as calcified coronary artery plaque. Image analysts are trained to reject calcification outside the anatomic boundary of the arteries or false-positive artifacts related to lymph nodes, pericardium, or motion. The review of candidate plaques by the image analyst is immediately repeated after image calibrations with the phantom data are performed. The software computes several measures of calcified coronary artery plaque, and these measures include the Agatston score (by using the standard 130-HU threshold, modified to adjust for section thickness), calcified plaque volume, and interpolated calcified plaque volume along the section direction. After image calibration, the measures of calcified plaque were recalculated to provide phantom-adjusted Agatston score, volume, and interpolated volume measures. The image analyst then completes a quality control assessment of quality in several categories: motion artifact, misregistration artifact, noise artifact, phantom placement, and coverage of the heart.
|
DATA TRANSMISSION AND ARCHIVING
|
|---|
The CT image data from the scanning sites were transferred electronically to the CT reading center by means of the Digital Imaging and Communication in Medicine (version 3.0) standard without data compression. Results of the analyses were electronically transferred from the central CT reading center to their respective data coordinating centers on a weekly basis. Coordinating centers monitored data flow to and from all sites and implemented computerized data tracking procedures to ensure timely, accurate, and complete data transmission. Image data and results were archived to compact disc and stored by the CT reading center. Backup copies were made and sent to the coordinating centers for off-site storage.
|
DATA QUALITY
|
|---|
Numerous procedures were designed and implemented to ensure high levels of CT image and data quality, facilitate protocol standardization across scanning sites, and reduce measurement error. Manuals in which the CT procedures, as well as the other examination components, are documented are publicly available and can be accessed through the Internet at www.cardia.dopm.uab.edu/ (CARDIA study) and www.mesa-nhlbi.org (MESA).
|
CT SCAN ACQUISITION QUALITY
|
|---|
Lead technologists at each imaging site participated in a 2-day training course that included didactic material, hands-on training, supervised training, and testing of proficiency. Full certification of a CT technologist was provided by the CT reading center only after successful completion and review of a series of pilot examinations. To confirm that technicians were obtaining scans in accordance with the protocol, the technical scanning parameters (tube voltage expressed in kilovolts, tube current–time product expressed in milliampere-seconds, etc) were extracted from the image headers by using an automated computer program for each participant’s scan. CT scans that deviated from the protocol were flagged for review by CT reading center personnel and communicated to the CT technologist and other scanning site personnel as appropriate.
After the image analyst analyzed each scan for calcified plaque, he or she completed a subjective assessment of image and scan acquisition quality in the following categories: motion artifact (defined as right coronary artery appearance as comma-shaped and twice its expected diameter or greater), misregistration artifact (defined as inconsistency in the anatomic borders of the heart displayed on sagittal or coronal reformatted images or between images of at least three transverse sections), noise artifact (defined as at least 30 false-positive lesions identified by scoring software on or near the coronary arteries), phantom placement (defined as cropping of portions of the calibration phantom from the image), and coverage of the heart (defined as failure to include portions of the coronary arteries). Images were rated as excellent or unacceptable in each category. Image quality scores for each CT technologist were reported regularly to the technologists and scanning site investigators to provide performance feedback, identify image quality problems, and direct additional training as necessary.
Acquisition of two consecutive CT scans in each MESA and CARDIA study participant yielded sequential measures of calcified coronary artery plaque for each participant. The rationale for obtaining replicate measures was to provide an improved point estimate of true calcified plaque burden, assess measurement error, and provide the opportunity for the CT technologist to correct any potential errors identified on the first scan (16–18). A physician investigator (R.C.D.) reviewed scans with the largest differences between the first and second scans to determine the reasons for the discrepancies (eg, misregistration) and correct obvious data errors.
|
QUALITY CONTROL OF CALCIFIED PLAQUE MEASUREMENT
|
|---|
To assess performance of the CT image analysts in the measurement of calcified coronary artery plaque, scans were randomly selected each quarter by the respective coordinating centers for inter- and intraobserver quality control reading (n = 806, MESA; n = 850, CARDIA study). A stratified sampling scheme that was based on the calcium score was used to select the scans for quality control readings, and target values at 70% (MESA) or 50% (CARDIA) of selected scans had a positive Agatston calcium score (a score > 0) for inter- and intraobserver analysis. The quality control examinations were selected during the performance of the examination, resulting in statistical fluctuations, with the actual values as follows: MESA—intraobserver reading, 240 (68%) of 351, and interobserver reading, 338 (74%) of 455; CARDIA—intraobserver reading, 230 (50%) of 460, and interobserver reading, 195 (50%) of 390. The initial image analyst reread half of the selected scans, and a different image analyst reread the other half to provide measures of intraobserver (within–image analyst) and interobserver (between–image analyst) variability. In addition, a fixed set of images was selected and reread at several time points throughout the examination period to assess temporal drift (change over time). The same physician investigator as mentioned previously reviewed the results of the quality control readings to determine potential reasons for observed discrepancies as part of the quality control process.
|
QUALITY CONTROL PHANTOM DATA
|
|---|
To monitor temporal drift within each CT system and to assess the comparability of the CT systems between scanning sites, a biweekly quality control scan was obtained. At each site, a standard torso insert positioned on the calibration phantom (Fig 2) was scanned, the torso and calibration phantom CT numbers were recorded, and these data were sent to the CT reading center. Calibrated torso values were calculated by linearly regressing the observed CT attenuation values (in Hounsfield units) from each of the calibration phantom regions on the image with the established values of calcium hydroxyapatite (0, 50, 100, and 200 mg/mL) contained in the phantom.
View larger version (144K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. Phantom used in MESA and the CARDIA study. The torso phantom is positioned on top of the quality control phantom containing the four cylinders with 0, 50, 100, and 200 mg/mL of calcium hydroxyapatite as part of the quality control procedures to track how the CT systems measure calcium with standardized conditions.
|
|
To correct for differences in CT image attenuation between scanners and over time, a calibration phantom was also used during participant scanning. Specifically, a calibration phantom (Image Analysis, Columbia, Ky) that contained three rods of a known density of calcium hydroxyapatite (50, 100, and 200 mg/mL) and a hydroxyapatite-free region (0 mg/mL) was placed underneath each participant during the CT examination and was imaged along with the participant’s heart in every acquired section (Fig 3). Measures of calcified plaque were determined with and without the phantom calibration.
View larger version (102K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3. Cardiac CT scan obtained in a study participant with quality control phantom in the transverse plane. The CT image demonstrates the quality control phantom positioned posteriorly to the supine participant during the cardiac CT examination. Four cylinders inside the phantom contain stable calcium concentrations of 0, 50, 100, and 200 mg/mL (from left to right) and were used to calibrate the CT numbers (Hounsfield units) for each scan. Note that the cylinder with 0 mg/mL of calcium is located on the far left and cannot be distinguished from the base material of the phantom.
|
|
|
CT EXAMINATION PARTICIPATION
|
|---|
Study participants successfully completed the CT examination at high rates in the CARDIA study and MESA. In the CARDIA study, at least one scan was obtained in 3044 (82.9%) of 3672 participants seen in the clinic, and in 3029 (82.5%) participants, complete data from paired scans were used for measuring calcified plaque. In the CARDIA study, participation in the CT examination was encouraged but not required for the year 15 examination (between May 2000 and September 2001). In 6814 (97.4%) of 6991 MESA participants seen in the clinic, complete data were on at least one scan, and in 6732 (96.3%) complete data were available on both scans. No complications or injuries were identified relative to the CT examination. Completion of a CT examination was a requirement for enrollment in MESA. A total of only 177 subjects were not enrolled after partial clinic visits; 28 of the 177 withdrew from the study, 62 were determined to be ineligible, and 87 were not enrolled for other reasons. In the CARDIA study, 629 participants who were seen in the clinic for the year 15 examination did not complete the CT component for the following reasons: 377 withdrew or did not return for a CT examination after the partial clinic visit, 43 were determined to be ineligible after a partial clinic visit, and 209 did not complete the CT component for other unspecified reasons.
|
CT IMAGE QUALITY
|
|---|
CT image quality was rated qualitatively by the image analysts during the reading process and was high for all sites and all CT technologies. Less than 1.0% of scans were rated as unacceptable with regard to streak artifact, phantom placement, and scan centering. Section registration or scan coverage was unacceptable for less than 5%. Image noise was rated as unacceptable in 6.2% and 11.7% of scans for MESA and the CARDIA study, respectively. Unacceptable image noise was higher for electron-beam CT than it was for four–detector row CT systems (8.3% vs 2.0% in MESA and 18.8% vs 4.3% in the CARDIA study, respectively). Unacceptable artifact secondary to motion was present in 7.7% of scans (measured in MESA only), and a higher percentage of unacceptable ratings was observed for four–detector row CT scanners than it was for electron-beam CT scanners (14.1% vs 4.5%).
|
STATISTICAL ANALYSES
|
|---|
Statistical analyses were performed as part of the quality control procedures. We specifically evaluated possible change over time (temporal drift) in the ability of the CT systems to measure calcium and how the image analysts compared in their performance of measuring calcified plaque with the CT scans and software at the central CT reading center (ie, intra- and interobserver agreement). Image analyst performance was measured in terms of agreement about the presence or absence of calcified plaque, as well as agreement about the amount of calcified plaque present.
|
CT SCANNER COMPARABILITY OVER TIME
|
|---|
The mean torso phantom values over time at each scanning site are presented in Table 2. All mean torso phantom values were within 2.7% of each other, and differences between scanning sites were not significant (P > .05). The variation in torso phantom values across time was less than 3% (standard deviation) for all sites, and the absence of a significant trend for any site over time (at the P = .05 level) indicated that the CT measurement of calcium was very stable over time at all sites in these studies.
|
AGREEMENT ABOUT PRESENCE OF CALCIFIED CORONARY ARTERY PLAQUE BETWEEN CONSECUTIVE CT SCANS
|
|---|
Presence or absence of calcified coronary artery plaque was highly comparable between the first and second scans obtained in each participant. Table 3 shows the proportion of participants who had no calcified plaque (Agatston score = 0) on both scans, those in whom both scans were positive for calcified plaque (Agatston score > 0), or those with a discordant scan set (ie, one scan that demonstrated calcified plaque and one that did not). In both MESA and the CARDIA study, observed agreement in regard to the presence of calcified coronary artery plaque on the consecutive CT scans for the same participant was 96%. The statistics of 0.92 (MESA) and 0.77 (CARDIA study) indicate strong agreement. The statistic is reduced for the CARDIA study, largely because of the lower prevalence (11.6% of CARDIA study and 50.2% of MESA participants had calcified plaque on at least one scan) and lower burden of calcified plaque found in the younger CARDIA study participants.
TOP
ABSTRACT
INTRODUCTION
