Janis Child

Comparison of coronary artery calcium scores between electron beam computed tomography and 64-multidetector computed tomographic scanner.

Objective: Because almost all data currently available with coronary calcium scanning are from electron beam tomography (EBT), we assessed whether scores obtained with 64-multidetector computed tomography (CT; MDCT) are similar. We evaluated the interscan variation in coronary artery calcium (CAC), Agatston score (AS), and volume score (VS) between EBT and 64-MDCT (VCT; GE, Milwaukee, Wis). Materials and Methods: One hundred two patients (mean age, 61.1 years; 27 women) underwent dual CAC scanning with both EBT and 64-MDCT. The AS and VS were measured with the Aquarius workstation (TeraRecon, Inc, San Mateo, Calif). The correlation coefficient, Bland-Altman analysis, interscanner variation, and agreement in AS and VS scores between EBT and 64-MDCT were computed. Results: Interscan agreement for presence of CAC was 99%. Median values were 286 and 268 mm2 for AS and 243 and 213 mm2 for VS with EBT and 64-MDCT, respectively (P > 0.05). There was significant linear relationship between scores from the 2 scanners (R = 0.98 in AS and R = 0.99 in VS; P < 0.001). The interscanner variability between EBT and 64-MDCT was 20.9% and 17.6% in AS and VS, respectively (P = NS). Bland-Altman analysis demonstrated a mean difference in scores of 8.3% for AS and 7.8% by VS. When compared with EBT, there were larger and more prevalent motion artifacts (P < 0.001) and larger mean Hounsfield units using 64-MDCT (P < 0.001). Conclusions: At CAC scanning, 64-MDCT and EBT were comparable in AS and VS. The interscan variability between scanners is similar to interscan variability of 2 calcium scores done on the same equipment. However, heart rate control was achieved for this study for calcium scores. Whether these results are repeatable without heart rate control needs to be further assessed.

Causes of interscan variability of coronary artery calcium measurements at electron-beam CT.

RATIONALE AND OBJECTIVES The authors performed this study to investigate the causes of interscan variability of coronary artery calcium measurements at electron-beam computed tomography (CT). MATERIALS AND METHODS Two sets of electron-beam CT scans were obtained in 298 consecutive patients who underwent electron-beam CT to screen for coronary artery calcium. Interscan variations of coronary artery calcium characteristics and the effects of heart rate, electrocardiographic (ECG) triggering method, image noise, and coronary motion on interscan variability were analyzed. RESULTS The interscan mean variabilities were 21.6% (median, 11.7%) and 17.8% (median, 10.8%) with the Agatston and volumetric score, respectively (P < .01). Variability decreased with increasing calcification score (34.6% for a score of 11-50 and 9.4% for a score of 400-1,000, P < .0001). The absolute difference in Agatston score between scans was 44.1 +/- 95.6. The correlation coefficient between the first and second sets of scans was 0.99 (P < .0001). Lower interscan variability was found in younger patients (<60 years), patients with stable heart rates (heart rate changing less than 10 beats per minute during scanning), patients with no visible coronary motion, and those with an optimal ECG triggering method (P < .05 for all). Results of multivariate logistic analysis showed that changes in calcium volume, mean attenuation, and peak attenuation were significant predictors of interscan variability and caused the interscan variations of the coronary artery calcium measurements (r2 = 0.83, P < .0001). CONCLUSION Coronary calcification at electron-beam CT varies from scan to scan. Volumetric scoring and optimal ECG triggering should be used to reduce interscan variability. Baseline calcium score and interscan variability must be considered in the evaluation of calcium progression.

Measurement of Thoracic Bone Mineral Density with Quantitative CT

PURPOSE To create standard thoracic bone mineral density (BMD) values for patients undergoing cardiac computed tomography (CT) by using thoracic quantitative CT and to compare these BMDs (in a subpopulation) with those obtained by using lumbar spine quantitative CT. MATERIALS AND METHODS The institutional review board approved this HIPAA-compliant study. A total of 9585 asymptomatic subjects (mean age, 56 years; age range, 30-90 years) who underwent coronary artery calcium scanning, including 4131 women, were examined. Patients with vertebral deformities or fractures were excluded. Six hundred forty-four subjects (322 of whom were female) also underwent lumbar quantitative CT. The mean thoracic vertebral BMDs for both sexes were reported separately in a subgroup of subjects aged 30 years and in 29 age-based subgroups in 2-year intervals from ages 30 to 90 years. The formulas used to calculate the female T score (T(f)) and the male T score (T(m)) on the basis of thoracic quantitative CT measurements were as follows: T(f) = (BMD(im) - 222)/36, and T(m) = (BMD(im) - 215)/33, where BMD(im) is the individual mean BMD. Comparisons between thoracic quantitative CT and lumbar quantitative CT measurements, as well as analyses of intraobserver, interobserver, and interscan variability, were performed. RESULTS The young-subgroup mean BMD was 221.9 mg/mL ± 36.2 (standard deviation) for the female subjects and 215.2 mg/mL ± 33.2 for the male subjects. The mean thoracic BMDs for the female and male subjects were found to be 20.7% higher and 17.0% higher, respectively, than the values measured with lumbar quantitative CT (P < .001 for both comparisons). A significant positive association between the thoracic and lumbar quantitative CT measurements (r > 0.85, P < .001) was found. Intraobserver, interobserver, and interscan variabilities in thoracic quantitative CT measurements were 2.5%, 2.6%, and 2.8%, respectively. CONCLUSION There was a significant association between the mean thoracic and lumbar BMDs. Therefore, standard derived measurements (young-subgroup BMD ± standard deviation) based on these data can be used with thoracic CT images to estimate the bone mineral status.

Improved accuracy of noninvasive electron beam coronary angiography.

Objectives:We investigated the effect of electrocardiographic (ECG) triggering on the accuracy of coronary electron-beam angiography (EBA) as compared with invasive angiography. Methods:One hundred thirty-three patients with suspected coronary disease were studied with intravenous coronary EBA and conventional coronary angiography. Patients were divided into 2 groups based upon ECG triggering on the EBA study. Patients were divided into 2 groups based upon different ECG triggering used: 80% R-R interval trigger method (group 1, n = 53) and end-systolic triggering (group 2, n = 80). End-systolic ECG triggering, which started at the end of the T wave in each study, was based on baseline heart rate. Results:Overall sensitivity to detect a ≥50% luminal stenosis was 69% in group 1 and 91% in group 2 (P = 0.002); specificity was 82% and 94% in group 1 and group 2, respectively (P < 0.001). Using newer triggering techniques (group 2) with EBA, the sensitivity, specificity, and accuracy for patients with disease of the left main coronary artery or 3 vessel disease was 100%, 94%, and 98%, respectively. Nonassessability of coronary segments on 3D-EBA images was reduced from 35% in group 1 to 9% in group 2 patients (P < 0.001). The number of motion-free coronary images increased from 67% to 95% from group 1 to group 2 (P < 0.0001). Conclusion:End-systolic ECG triggering improves accuracy, image quality, and assessability of segments of coronary EBA for the detection of angiographic coronary artery disease.

Sensitivity to detect small coronary artery calcium lesions with varying slice thickness using electron beam tomography.

Mao S, Child J, Carson S, et al. Sensitivity to detect small coronary artery calcium lesions with varying slice thickness using electron beam tomography. Invest Radiol 2003;38:183–187. Rationale and Objective. To estimate the sensitivity to find small coronary artery calcium lesions with use of different slice widths with electron beam tomography. Materials and Methods. Two studies were performed. Study 1 utilized double scanning of a stationary cork phantom with three different slice thickness (1.5, 3, and 6 mm). Fifty different calcific lesions (all <20 mm2 in area) fitted in 10 cork coronary arteries were utilized. The calcium foci area, peak value and score were measured and compared. In group 2, 30 patients underwent coronary artery calcium (CAC) screen studies. Each patient was scanned with both 3-mm and 6-mm scan widths in a same study time. Lesions with < 20 mm2 of area of CAC were measured on both 3-mm and 6-mm images. The mean and peak Hounsfield unit measure, and Agatston score were compared between both images. Results. In the cork study, the sensitivity to detect small calcium foci were 96% (48/50), 82% (41/50), and 34% (17/50) in images with 1.5-, 3-, and 6-mm slice thickness, respectively. There is a smaller value in mass, and calcium volume in 6-mm images than 1.5-mm and 3-mm images (P < 0.001). There was no significant difference between the true value and measured value from 1.5-mm and 3-mm images. In the human study, 18 (30%) of 60 CAC lesions with an area < 20 mm2 defined on 3 mm images were not visible on 6-mm images. Sensitivity of small lesions (< 5 mm2) was 48% using 6-mm slices. There was a smaller value in CAC area, mean and peak Hounsfield units and score measured from 6-mm images, as compared with 3 mm slices (P < 0.05). Conclusion. Thinner slice imaging has a higher sensitivity to detect small calcium focus. There was no significant change in score between 3 mm and 1.5 mm on the cork phantom study. However, the use of 6-mm slices should be discouraged, as this protocol both underestimates calcific mass and misses a significant number of calcific lesions in both a phantom and human study.

Trabecular Bone Mineral Density Measurement Using Thoracic and Lumbar Quantitative Computed Tomography

PURPOSE To evaluate the agreement of bone mineral density (BMD) between lumbar (L) and individual thoracic (T) vertebrae and identify a standard thoracic spine level for BMD assessment in cardiac computed tomography (CT) images. MATERIALS AND METHODS Three hundred subjects who underwent simultaneous chest and abdomen CT scans for clinical indications were included. A calibration phantom that extended from the first thoracic spine (T(1)) to the fifth lumbar (L(5)) was employed. Vertebral BMD were measured by QCT 5000 and NVivo systems. The association between three consecutive lumbar (L1-L3) and thoracic BMD (3T, initiation site equivalent to left main coronary caudally) was evaluated. RESULTS There was a gradual decrease in BMD values from T(1) to L(3,) subsequently increasing in L(4) and L(5) in both genders. When stratified by gender, 3T BMD was significantly higher versus L(1-3) BMD (156.9 versus 141.9vmg/cm(3), P < .001) for women as well as for men (164.8 versus 151.0 mg/cm(3), P < .001). There is good correlation between 3T and L(1-3) BMD, the Pearsons correlation coefficients are 0.91 and 0.93 for women and men, respectively. We further analyzed the associations between L(1-3) and any individual spine of T(1)-L(5) and similar relationships were observed (r value, 0.62-0.98). The intraobserver, interobserver, and interscan variation measurement of thoracic quantitative CT was 2.5 (1.0, 95% CI 0.099-1.004); 2.6 (1.0, 95CI% 0.992-1.007), and 2.8% (1.0,95% 0.0994-1.008), respectively. CONCLUSION The 3T BMD was highly correlated with L(1-3) BMD. Thoracic BMD can be measured during cardiac and lung CT imaging without need for additional participant burden or radiation dose. This highly reproducible methodology is actively being applied to large cohort studies to evaluate the prevalence of osteoporosis and track BMD over time.

Comparison of LV mass and volume measurements derived from electron beam tomography using cine imaging and angiographic imaging.

Purpose: To estimate the variation of left ventricular (LV) mass and volume measurement with cine and angiography by electron beam tomography (EBT). Method and materials: Sixty-three consecutive patients (41 men, 22 women; age range 46–91) referred for cardiac imaging for clinical indications underwent cine and coronary artery electron beam angiography (EBA) studies on the same day. The cine images consisted of 144 images (12 slices/level × 12 levels), taken 12 frames/s for a full cardiac cycle. The EBA images consisted of 50–70 slices triggered at end-systole, with an acquisition time of 100 ms/slice. Slice thickness was 8 mm for the cine images and 1.5 mm for the EBA images. A total volume of 120–180 ml of nonionic contrast was used for each subject. The LV mass (myocardial tissue volume), LV cavity volume and total LV volume (tissue + cavity) measurements were completed using the software from the EBT computer console (G.E., S. San Francisco, CA). Results: The LV mass, cavity volume and total LV volumes at end-systole were 124.11 g, 45.66 and 163.86 ml when derived from the cine images and 130.74 g, 41.31 and 165.82 ml when derived from the EBA images. There were no significant differences between the cine and EBA-derived measurements, however the EBA-derived measurements showed slightly larger LV mass (mean 6.63 g), smaller cavity volume (mean −4.35 ml) and larger total LV volume (mean 1.96 ml, all p > 0.05) than did the cine-derived measurements. Based on case-by-case observations, these differences appear to be related to the higher spatial resolution of the thinner EBA images which allows better discrimination between papillary and trabecular muscle and LV. This leads to slightly smaller cavity size estimations and greater LV mass measurements. There was significant correlation between cine and EBA-derived measurements. Formulas were developed for relating the measurements made from the two modalities as follows: For LV mass: EBA value = 0.91 × cine value + 17.09, R = 0.95, p < 0.001; For LV cavity volume: EBA value = 1.06 × cine value − 6.91, R = 0.96, p < 0.001; For total LV volume: EBA value = 0.98 × cine value + 5.09 in ml, p < 0.001. The mean differences in measurements using the two modalities were 8.1, 18.2 and 6.5% for LV mass, LV cavity volume and total LV volume, respectively. Conclusion: Both cine and EBA images were useful for measuring LV mass and volume with good inter-test agreement. Cardiac volume and mass measurements derived from cine EBT studies probably slightly underestimate LV mass and overestimate LV volume.

Ability of calibration phantom to reduce the interscan variability in electron beam computed tomography.

Objective To test the hypothesis that a calibration phantom would improve interpatient and interscan variability in coronary artery calcium (CAC) studies. Methods We scanned 144 patients twice with or without the calibration phantom and then scanned 93 patients with a single calcific lesion twice and, finally, scanned a cork heart with calcific foci. Results There were no linear correlations in computed tomography Hounsfield unit (CT HU) and CT HU interscan variation between blood pool and phantom plugs at any slice level in patient groups (p > 0.05). The CT HU interscan variation in phantom plugs (2.11 HU) was less than that of the blood pool (3.47 HU; p < 0.05) and CAC lesion (20.39; p < 0.001). Comparing images with and without a calibration phantom, there was a significant decrease in CT HU as well as an increase in noise and peak values in patient studies and the cork phantom study. Conclusion The CT HU attenuation variations of the interpatient and interscan blood pool, calibration phantom plug, and cork coronary arteries were not parallel. Therefore, the ability to adjust the CT HU variation of calcific lesions by a calibration phantom is problematic and may worsen the problem.

Coronary calcium test phantom containing true CaHA microspheres for evaluation of advanced CT calcium scoring methods

BACKGROUND Test phantoms with simulated micro-calcifications of true calcium hydroxyapatite (CaHA) density were not available to validate advanced calcium scoring methods or plaque density measurements. OBJECTIVES We evaluated a coronary calcium scoring (CCS) test phantom containing very small CaHA microspheres and validated a new scoring method for measurements of plaque densities. METHODS The semianthropomorphic CCS phantom was constructed with CaHA microspheres (volumes, 0.05-3.1 mm(3)) with the approximate density of biologic calcifications. QRM and CCS phantoms were scored with a new calibrated and automated calcium scoring method (N-vivo; Image Analysis). The densities of the microspheres and 609 individual patient plaques were measured. RESULTS The range of measured densities of the CaHA microspheres was approximately equivalent to that measured in the patient coronary calcifications. The smallest microspheres scored with the calibrated/automated and the Agatston methods had volumes of 0.075 mm(3) and 0.27 mm(3), respectively. The standard deviations of the mass scores of the microspheres ranged from 0.02 to 0.17 mg with regression slope of 0.962 and R(2) = 0.997. The relationship of measured density to measured mass of the patient plaques was similar to that of the microspheres, suggesting that vascular calcifications are CaHA density. CONCLUSIONS The CaHA microspheres of the CCS test phantom were found to be representative in density and size of coronary calcifications. The measurements show that CT calcium scoring underestimates plaque density and greatly overestimates volume. The heterogeneity of calcium concentration densities measured in the patient plaques was due largely to CT scanner measurement errors.