Author + information
- Received September 26, 2011
- Revision received December 8, 2011
- Accepted December 22, 2011
- Published online June 1, 2012.
- Kambiz Ghafourian, MD, MPH⁎,
- Desiree Younes, MD†,
- Lauren A. Simprini, MD⁎,
- Wm. Guy Weigold, MD⁎,
- Gaby Weissman, MD⁎ and
- Allen J. Taylor, MD†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Allen J. Taylor, Cardiology Division, Department of Medicine, Georgetown University Hospital, 3800 Reservoir Road, NW, Washington, DC 20007
Objectives The study evaluated the relationship between cardiac computed tomography (CT) scout view x-ray attenuation and CT image noise compared with weight or body mass index (BMI).
Background Decreasing peak tube voltage from 120 to 100 kVp on the basis of body size reduces radiation exposure. Methods to better predict CT image noise may lead to more effective selection of reduced tube voltage in cardiac CT.
Methods Image quality was graded subjectively (1 [excellent] to 4 [nondiagnostic]) and objectively (SD of the aortic attenuation value) in cardiac CT angiograms (N = 106) acquired at either 100 or 120 kVp. X-ray attenuation characteristics on the scout view (120 kVp, 30 mA) were measured within a 3-cm region of interest across the chest in the frontal x-ray. Receiver-operating characteristic curve analysis was performed comparing scout view attenuation versus weight and BMI in predicting CT image noise and quality.
Results CT image noise correlated with both BMI (r = 0.40; p < 0.001) and the scout view attenuation value (r = 0.52; p < 0.001). In linear regression models with controlling for BMI (or weight) and tube potential, scout view attenuation was the best predictor of the CT image noise (p < 0.001), and increased model fit statistic from 0.23 to 0.41 (p model <0.001). At 120 kVp, scout view attenuation predicted CT image noise <30 Hounsfield units (HU) more accurately than BMI (area under the curve: 0.89 vs. 0.77). For CT images acquired at 120 kVp, those with a scout view attenuation <−120 HU had significantly lower noise and higher signal-to-noise ratios, with similar mean aortic attenuation values. A majority (89.3%) of “low-noise” CT images at 120 kVp had scout view attenuation values of <−120 HU.
Conclusions Scout view attenuation predicts cardiac CT image noise better than weight or BMI and could enable broader application of reduced x-ray tube voltage as a radiation sparing technique.
Cardiac computed tomography (CT) angiography provides a noninvasive, accurate method to evaluate diverse aspects of cardiac structure and function (1). Optimal use of the test requires a tradeoff between the potential risks of radiation exposure (2) and the diagnostic quality of the images (3). Such risks may be amplified in younger patients and women (4), and in those with high levels of exposure (5). Methods are needed to improve or preserve image quality, while at the same time reducing radiation exposure.
Radiation dose sparing scanning techniques may adversely affect image quality due to the inverse relationship between radiation exposure and image noise. Image noise, quantified as the standard deviation of CT attenuation values within a uniform region of interest, decreases as the number of photons received by the detector array increases (6). The number of photons received is dependent on the photon output produced by the x-ray tube (determined by tube potential, tube current, and exposure time), and the degree of x-ray attenuation exerted by the patient's body. Thus, there is a tradeoff between image noise and radiation exposure to the patient.
Tube potential reductions can provide significant radiation dose sparing. Decreasing peak tube potential from 120 to 100 kVp has been shown to reduce effective radiation dose by up to 46% while simultaneously preserving image quality in appropriately selected patients (7,8). Presently, empiric selection criteria for application of reduced peak tube potential include markers of body size (e.g., weight <85 kg or body mass index [BMI] <30 kg/m2) because increased x-ray attenuation can increase image noise at these lower photon energies. Body size selection criteria are likely crude given interpatient differences in body composition and weight distribution. Thus, an alternative approach would be to measure body attenuation characteristics during the planning phase of the CT examination. Prior to CT imaging, patients undergo a scout x-ray view, similar to a routine planar x-ray, to define the CT scan range. We hypothesized that attenuation data derived from the scout view would correlate more strongly with both objective and subjective measures of CT image quality than either BMI or weight, and thereby lead to more informed tube potential settings during cardiac CT.
Study population and data acquisition
A consecutive series of contrast enhanced cardiac CT angiograms performed at Washington Hospital Center during a 2-month period were reviewed. Cases not eligible for inclusion included those with missing data for height and weight (n = 7), and CT examinations performed at peak tube potential of 140 kVp (n = 5) or those in which the scout view were acquired using tube settings other than 120 kVp and 30 mA (n = 2). All patients were scanned with a 256-slice multidetector CT scanner (Brilliance iCT, Philips Healthcare, Cleveland, Ohio). Iodinated contrast agent (Iohexol 350, GE HealthCare, Princeton, New Jersey) was used in a triphasic injection protocol, which consisted of an initial injection of 50-ml full-strength contrast at 4 to 5 ml/s, followed by 10 ml of contrast at 2 to 2.5 ml/s, followed by 30 to 40 ml of saline at 4 to 5 ml/s. Accurate timing of the scan was determined by bolus tracking with automatic scan triggering when the density in the region of interest (ROI) increased to the preset value of 175 Hounsfield units (HU). The ROI was located in the main pulmonary artery with a 12- to 14-s delay for axial acquisitions and in the descending aorta with a 9-s delay for the helical acquisition. All patients received intravenous or oral beta-blocker to achieve heart rate of <65 beats/min. In the absence of contraindication, patients also received pretreatment with nitroglycerin (400 to 800 μg sublingual).
A scout scan of the chest in the anterior-posterior projection was used in all cases to determine the minimum range of the cardiac CT examination. All scout scans were performed at a routine setting of 120 kVp and 30 mA. Peak tube potential was set at either 120 or 100 kVp on the basis of the patient's weight or body habitus; 120 kVp was used as a default setting, but the tube potential was reduced to 100 kVp when the patient's weight was below 85 Kg or the BMI was below 30 kg/m2.
Cardiac CT examinations selected for this analysis were required to meet the following characteristics: 1) scout scans obtained using scanner settings of 120 kVp and 30 mA; 2) peak tube potential of either 120 or 100 kVp; and 3) CT images with image reconstruction at the 75% phase of the cardiac cycle. Slice width was 0.9 mm in all the included images. For filtered back projection, we used a standard reconstruction filter (XCB) in this study. The clinical CT imaging database of the Washington Hospital Center was used to extract the following variables: height; weight; gender; age; scout view x-ray tube current (mA); tube current (mA); tube potential (kVp); and the acquisition mode (helical vs. axial). Cardiac CT examinations without data for weight and height were excluded. The study was approved by the institutional review board at the Washington Hospital Center.
Scout scan assessment
From the scout anterior-posterior view, a rectangular ROI of 3 cm in height was drawn across the chest, 1 cm inferior to the bifurcation of the trachea, as demonstrated in Figure 1. The mean x-ray attenuation of the scout view was then determined by the average of 2 repeated measurements of the mean attenuation.
Subjective image quality
For each participant, subjective quality of the CT angiogram was graded by 2 reviewers according to a 4-point visual scale of 1 (excellent quality) to 4 (nondiagnostic quality) as described in Figure 2. We defined a “good-quality” image as a subjective score of ≤2. Image quality assessment was performed using standardized display settings with a center of 200 HU and width of 800 HU.
Objective image quality (image noise)
The mean ± SD CT attenuation (HU) were measured within a 1-cm2 circular ROI placed within the aortic root at the level of the pulmonary artery bifurcation for 3 consecutive slices at the 75% phase. The image noise was then derived from the average of the 3 measurements of the SD values. We define a “low-noise” image as mean ± SD <30 HU at the aortic ROI.
The distribution of continuous variables was tested for normality by the Shapiro-Wilk normality test. Continuous variables are presented as mean ± SD or median ± interquartile range. Categorical variables are presented as absolute and relative frequencies (%). The relationship between the subjective and objective image quality was evaluated using analysis of variance. Pearson correlation coefficients were used to evaluate the correlation between the objective CT image noise and both BMI and scout view attenuation at 100 and 120 kVp. We used multivariate linear regression analysis to explore the relationship between scout view attenuation and CT image quality at 100 and 120 kVp. All the variables that were included in the model fulfilled the linear regression assumptions. Receiver-operating-characteristic (ROC) curve analysis was performed to compare the BMI with the scout view attenuation in predicting “low-noise” image quality, which was defined as an aortic SD <30 HU similar to the threshold used by others (9). We also used ROC curve analysis to assess the sensitivity and specificity of the scout view attenuation in predicting “low-noise” image quality. All reported p values were based on 2-sided tests and were compared to a significant level of p < 0.05. All data were analyzed using SPSS version 16.0 for Windows (IBM, Chicago, Illinois).
Characteristics of the study population and indications for the cardiac CT examination are presented in Tables 1 and 2.⇓⇓ Mean body weight was 83.9 ± 16 kg and mean BMI was 28.4 ± 5.3 kg/m2. Peak tube potential was set at 120 kVp in 64 cases (60.4%) and 100 kVp in the other 42 (39.6%). The majority of the scans were acquired with a prospectively electrocardiography (ECG)-triggered sequential scan protocol (axial CT), whereas retrospectively ECG-gated helical data acquisition was used in 17.9% of the cases. The mean heart rate was 67 ± 20 beats/min.
Results of the CT image quality, scout view evaluation, and the radiation dose are presented in Table 3. In the subjective assessment of the CT images, 84 cases (79.1%) had “excellent” or “good” image quality (intraclass correlation coefficient 0.69, p < 0.001). The median CT image noise was 32.8 HU for the study population (interquartile range: 27.8 to 39.1 HU). There was a statistically significant direct correlation between the subjective image quality and the objective assessment of the CT image noise (analysis of variance p < 0.001). BMI and scout view attenuation showed significant bivariate correlations with the CT image noise (r = 0.52, p < 0.001; and r = 0.40, p < 0.001, respectively). After adjustment for BMI and tube potential in a linear regression model, scout view attenuation was the best predictor of the CT image noise (p < 0.001). Inclusion of the scout view attenuation to the model, which included BMI and tube potential, improved the model fit statistic (R2) from 0.23 to 0.41 (p model <0.001). The regression equation for image noise was [noise = 62.6 + (0.73 × BMI) – (10.2 × kVp) + (0.7 × scout attenuation)]. After adjustment for the scout view attenuation, BMI was not statistically significantly correlated with the CT image noise (p = 0.79).
“Low-noise” CT image quality was most accurately predicted by scout view attenuation (ROC area under the curve [AUC]: 0.73; 95% confidence interval [CI]: 0.63 to 0.82; p < 0.001), followed by BMI (ROC AUC: 0.67; 95% CI: 0.56 to 0.78; p = 0.004) and body weight (ROC AUC: 0.57; 95% CI: 0.46 to 0.68; p = 0.21). At 120 kVp, all 3 variables significantly predicted image quality, although the highest AUCs were observed for scout view density. The AUC for scout view density, BMI, and weight were 0.89 (95% CI: 0.81 to 0.97; p < 0.001), 0.77 (95% CI: 0.66 to 0.89; p < 0.001), and 0.66 (95% CI: 0.53 to 0.80; p = 0.03), respectively. At 120 kVp, scout view attenuation was superior to body weight (p < 0.05), but not significantly different than BMI. At 100 kVp, scout view density, BMI, and weight had similar predictive value for image noise (AUC: 0.80, 0.78, and 0.77, respectively; p < 0.01 for all). Scout view attenuation more accurately predicted “low-noise” image quality at 120 kVp (AUC: 0.89) (Fig. 3) than 100 kVp (AUC: 0.80). For CT images acquired at 120 kVp, a cutoff value of −120 HU for the scout view attenuation value was found in 37 of 62 patients (59.7%) and had a 67.6% sensitivity and 88% specificity to predict a CT image noise <30 HU (Table 4). Among those imaged at 120 kVp, those with a scout view attenuation <−120 HU had significantly lower noise (28 ± 10 HU vs. 44 ± 13 HU; p < 0.001) and higher signal-to-noise ratios (12.5 ± 3.9 vs. 7.6 ± 2.7; p < 0.001), with similar mean aortic attenuation values (329 ± 75 HU vs. 311 ± 65 HU; p = 0.33). In this study, 89.3% of the “low-noise” CT images at the peak tube potential of 120 kVp had scout view attenuation values of <−120 HU (Table 4).
Despite substantial progress in the development of radiation-dose sparing techniques in cardiac CT angiography that will minimize the potential long-term adverse effects of radiation exposure, image quality must be maintained to preserve diagnostic value of the images. Because of the inverse relationship between x-ray attenuation and image quality, objective assessment of x-ray attenuation may be useful in identifying individuals in whom radiation dose can be decreased without clinically significant effects on image quality. The present study demonstrates that scout view attenuation correlates more strongly with image noise than BMI and body weight. A scout view attenuation <–120 HU was a specific predictor of image noise <30 HU.
A number of scanning techniques using individualized parameters have been evaluated and shown to be effective at decreasing radiation exposure without affecting image quality such as minimizing scan length and the duration of radiation exposure (7,10). ECG-controlled tube current modulation can decrease radiation exposure by 25% to 50% (10) and is most effective at lower heart rates. In patients with regular and slower heart rates, more radiation-dose sparing can be achieved with prospective ECG-gating acquisition, which decreases radiation exposure by 80% compared with retrospective ECG-gating (11,12).
Recent data support imaging with reduced peak tube potential of 100 kVp rather than 120 kVp, leading to a substantial reduction in radiation exposure (7,8) including during helical data acquisition with tube current modulation (10). In the Michigan Blue Cross Blue Shield registry, average radiation exposure was recorded in 15 centers before and after the implementation of a best-practice cardiac CT model. This study identified that the increased use of 100 kVp imaging from 13% (pre-intervention) to 43% (post-intervention) was the most important acquisition parameter associated with achieving a target radiation dose of <15 mSv (13). Thus, broader application of 100 kVp setting creates the potential for effective and meaningful reductions in radiation exposure.
In our study, similar to that of Raff et al. (13), 100 kvp imaging could be applied to only 40% of patients when selected according to body size characteristics. However, among those scanned at 120 kvp, a majority had mean scout x-ray attenuation values of <–120 HU, a threshold associated with “low-noise” CT image quality. Using this method, reduced tube potential may be more broadly applied without adversely affecting diagnostic image quality. These results are consistent with another study showing a linear relationship between scout view image attenuation and image noise in cardiac CT images at 120 kvp (9).
In this study, we defined acceptable objective image quality as a noise value of <30 HU (9). However, image quality exists as a continuum, and this threshold does not alone indicate that studies above this threshold would necessarily have limited diagnostic accuracy. The method studied for measurement of scout view x-ray attenuation was a manual measurement of mean attenuation of a 3-cm region of interest across the chest. Prospective validation of this method and resultant acquisition protocols is required, including broader evaluation of the impact of scout view x-ray attenuation assessments in individuals with low body weights to confirm the appropriateness of 100 kVp imaging and potentially select even lower tube potential settings. Furthermore, although simple in its application, methods to automate such assessments could promote greater use. These results are applicable only to CT images reconstructed using filtered back-projection and newer image reconstruction methods such as iterative reconstruction algorithms may show different relationships between scout view attenuation and CT image noise.
Scout view attenuation values correlate more closely with CT image noise than either weight or BMI, and thus may be useful in more broadly identifying individuals in whom imaging with reduced tube potential could lead to lower radiation exposure with preserved image quality.
All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- area under the curve
- body mass index
- confidence interval
- computed tomography
- Hounsfield unit
- receiver-operating characteristic
- region of interest
- Received September 26, 2011.
- Revision received December 8, 2011.
- Accepted December 22, 2011.
- American College of Cardiology Foundation
- Taylor A.J.Cerqueira M.,
- Hodgson J.M.,
- et al.
- Budoff M.J.,
- Achenbach S.,
- Blumenthal R.S.,
- et al.
- Gerber T.C.,
- Carr J.J.,
- Arai A.E.,
- et al.
- Morin R.L.,
- Gerber T.C.,
- McCollough C.H.
- Hausleiter J.,
- Martinoff S.,
- Hadamitzky M.,
- et al.
- Hausleiter J.,
- Meyer T.,
- Hadamitzky M.,
- et al.