The human research committee of our institutional review board approved the research study with a waiver of informed consent. The study was performed in compliance with the guidelines established by the Health Insurance Privacy and Portability Act (HIPPA). The study cohort comprised 152 patients (male-female ratio, 78:74; mean age, 60 years; age range, 25-101 years) who underwent contrast-enhanced CT examination of the abdomen and pelvis with a 16-MDCT scanner (Somatom Sensation 16, Siemens Medical Solutions). All studies were clinically indicated for assessing abdominal and pelvic diseases. Of these 152 patients, 79 consecutive patients were scanned with the combined or xyz-axes modulation technique (CARE Dose 4D, Siemens) using either weak decrease (slim)-strong increase (obese) type (n = 42; male-female ratio, 21:21; mean age, 61 years; age range, 25-84 years) or average decrease (slim)-average increase (obese) type (n = 37; male-female ratio, 20:17; mean age, 63 years; age range, 35-83 years). An image quality reference of 160 mAs was used for scanning the patients in the groups. The effective milliampere-second setting can be defined as the tube current-time product divided by the pitch factor.

View larger version (46K)

Fig. 1 —Line graph shows relationship between attenuation profile and effective milliampere-second (mAs) setting. Adaptation strength can be adjusted separately for left branch (slim patients) and right branch (obese patients) of curve. White line indicates theoretic limits of adaptation for constant image noise. Absolute effective mAs value is scaled with image quality reference mAs value selected for performing CT study.

Of the remaining 73 patients, 42 consecutive patients (male-female ratio, 23:19; mean age, 61 years; age range, 25-94 years) were scanned using only the angular modulation technique (CARE Dose 4D, Siemens) with an effective milliampere-second setting of 160-200. Thirty-one consecutive patients were scanned with a fixed or constant tube current (effective mAs, 160-200; male-female ratio, 14:17; mean age, 60 years; age range, 25-101 years). Whereas the user needs to specify a milliampere value for scanning on CT systems manufactured by other vendors, the effective milliampere-second value is used to specify a desired tube current to perform scanning with the constant tube current or angular modulation technique on the CT scanner used in our study. “Effective mAs” is a vendor-specific term that has been defined as the tube current-gantry rotation time product (mAs) divided by pitch. Weights of all patients were recorded before the CT examinations. The mean weight of the study cohort was 77.2 kg, with a range of 41-136 kg.

To facilitate comparison of the radiation dose, the remaining scanning parameters were kept identical for each patient regardless of the tube current technique used. These scanning parameters included 140 kVp, 0.5-sec gantry rotation time, 16 × 1.5 mm detector configuration, 24-mm table feed per gantry rotation, 5-mm reconstructed slice thickness, 5-mm slice interval, and B31 medium soft-tissue reconstruction algorithm. In addition, the CT radiation dose descriptors—that is, CT dose index volume (CTDI_vol) and dose-length product (DLP)—were recorded for each study. Because tube current changes during each gantry rotation and at each slice position with automatic tube current modulation techniques, the system takes into account the modulation of exposure and automatically provides the average CTDI_vol and aggregate DLP values on the user interface for studies performed with modulation techniques.

The online real-time, anatomy-adapted, attenuation-based angular modulation technique (CARE Dose 4D) does not need information from a radiographic localizer to adapt the tube current. In noncircular cross-sectional regions, the X-ray beam attenuation varies in different projections, and in these settings, image noise in high beam attenuation projection angles determines the overall image noise content. Without angular modulation, the tube current remains constant over 360° rotation regardless of regional attenuation. With angular modulation, the tube current is reduced as a function of the projection angles for low-attenuation projections (reduced tube current in anteroposterior projection compared with that in lateral direction). The angular modulation technique assessed in our study estimates the modulation function data from the real-time regional attenuation profile. These data are analyzed and relayed to the generator control for adapting tube current with a 180° delay from the X-ray generation angle. In asymmetric regions, such as the shoulders, with significantly less beam attenuation in the anteroposterior direction than in the lateral direction, a substantial reduction in tube current occurs in the anteroposterior or posteroanterior direction with the use of angular modulation.

As stated in the previous section, the combined modulation technique used in the present study includes automatic tube current adaptation in the scanning direction (z-axis modulation, which determines a tube current value for each slice position in the z-axis) and online tube current modulation for each tube rotation (angular modulation, which uses a different tube current value for different beam projections at each slice position). For the z-axis modulation component of the combined modulation technique, an attenuation profile along the patient's long axis (z-axis) is measured in the direction of the projection and is estimated for the perpendicular direction with a mathematic algorithm on the basis of a single localizer radiograph. The attenuation profile consists of information regarding the patient's size, anatomic shape, and density at each position in the z-axis. On the basis of these attenuation profiles, axial tube current values are calculated to adapt tube current for z-axis modulation. An analytic function defines the correlation between attenuation profile and tube current for slice position in the z-axis and adapts the tube current to patient size and attenuation changes. Tube current adjustment is based on a user-defined image quality reference milliampere-second setting to maintain the desired image quality in all images along the scanning direction (z-axis modulation component).

On the basis of these levels, the technique also modulates the tube current online during each tube rotation according to the patient's angular attenuation profile (angular modulation component). The image quality reference mAs value is selected according to the diagnostic requirements and the preference of the radiologist. For a given scanning protocol, this value reflects the effective mAs that is used for a reference patient defined as a typical adult weighing 70-80 kg (for adult protocols) or a typical child weighing 20 kg (for pediatric protocols).

The combined modulation technique adapts tube current to the size of the individual patient on the basis of the image quality reference mAs value, which is changed only if an adjustment to image quality is required and not for individual patient size (Fig. 1). The use of different modulation settings for slim and obese patients, as seen in Figure 1, is supported by prior studies of the z-axis modulation technique (not assessed in the present study) that have reported a substantial difference in image quality with the use of similar tube current modulation settings in patients of different sizes [3]. The technique decreases tube current for slim patients and increases tube current for obese patients. However, the extent of change in tube current can be controlled with the use of the appropriate modulation strengths, which may be set as weak, average, or strong. Thus, the strong modulation setting for obese patients results in a larger increase in radiation dose and in reduced image noise than the preset image quality reference mAs and average modulation setting. The weak modulation setting for obese patients results in more image noise and lower radiation dose than the preset image quality reference mAs and average modulation setting. Conversely, compared with the preset mAs, a strong modulation setting for slim patients results in more image noise and lower radiation dose. Likewise, a weak modulation setting for slim patients results in less image noise and higher radiation dose. These modulation types for slim and obese patients were determined from preclinical studies conducted by the vendor (unpublished data, Siemens Medical Solutions).

To use the combined modulation technique, the modulation strength for slim and obese patients must be selected before acquisition of the localizer radiograph; the modulation strength can be constant for all patients or can be changed on the basis of the diagnostic requirement for image quality. The software determines whether a patient is slim or obese from the localizer radiograph and modulates the dose based on the preselected modulation strength for these patients. Image quality and radiation dose can be controlled by selecting an appropriate setting of combined modulation and image quality reference mAs value.

Quantitative image noise was recorded for each examination in the liver parenchyma at the level of the porta hepatis with a region of interest of constant rectangular shape and 30 square pixels. Each region of interest was placed in a homogeneous region in the liver parenchyma without obvious vessels or focal liver lesions.

Two subspecialty radiologists, one with 8 years' experience and the other with 3 years' experience, blinded to the scanning techniques, independently reviewed each study on a digital PACS diagnostic workstation (AGFA Impax RS 3000 1K review station, AGFA Technical Imaging Systems). All CT examinations were assessed at the same window level and width level of 40 and 400 H, respectively. Each study was assessed independently for image noise, diagnostic acceptability, presence of streak artifacts, and visibility of small structures at three different levels in the abdomen and pelvis. These subjective image quality parameters were selected on the basis of prior studies [4] and European Guidelines on Quality Criteria for CT recommendations [5].

Each radiologist graded image noise using the following 5-point scale: 1, very little noise; 2, better than average noise; 3, acceptable noise; 4, more than acceptable noise; and 5, too much noise. Image noise was categorized as acceptable if there was average mottle or graininess with acceptable visualization of anatomic structures and interfaces between structures of different attenuation. Too much noise was defined as graininess or mottle that interfered with visualization of these structures, and very little noise was graded on the basis of minimal image graininess.

Diagnostic acceptability was graded using the following 4-point scale: 1, fully acceptable; 2, probably acceptable: 3, only acceptable in limited conditions; and 4, unacceptable. Fully acceptable diagnostic acceptability was defined as acceptable contrast, sharpness of different structures, and lesion visualization. Unacceptable diagnostic acceptability was defined as completely unsatisfactory visualization of these image attributes. Visibility of small structures, such as small blood vessels, lymph nodes, or adrenal glands, was graded on a 5-point scale: 1, excellent; 2, above average; 3, acceptable; 4, suboptimal; and 5, very poor. Streak artifacts were graded using a 3-point scale: 1, absent; 2, present without affecting visibility; and 3, present and affecting visibility.

The Student's t test (Excel, Microsoft) was performed to compare the ages and weights of patients who were scanned with any of the three techniques (combined modulation, angular modulation, and constant tube current). The Wilcoxon's signed rank test (MedCalc software, MedCalc) was used to compare image noise, diagnostic acceptability, presence of streak artifacts, and visibility of small structures for examinations performed with the different techniques. To minimize chance occurrence of significant statistical difference with multiple comparisons (four qualitative parameters × three levels = 12 comparisons), Bonferroni correction [6] was applied to find the appropriate level of alpha or p value for statistical significance. The alpha level for significant statistical difference for each test was lowered to 0.0042 to bring the alpha level overall back to 0.05 for multiple comparisons performed in the present study.

Means and SDs of quantitative noise, CTDI_vol, and DLP were determined for examinations performed with the combined modulation, angular modulation, and constant tube current techniques (Excel). Quantitative image noise values for examinations performed with different techniques were compared with the analysis of variance (MedCalc software). Likewise, CTDI_vol and DLP for examinations performed with the aforementioned techniques were also compared using the analysis of variance test. Linear statistical correlations between patient weight and CTDI_vol for examinations performed with the different techniques were calculated using MedCalc software. The degree of interobserver concordance was determined using the kappa test. Kappa coefficient values for interobserver agreement were considered as slight if less than 0.2; fair, 0.21-0.40; moderate, 0.41-0.60; substantial, 0.61-0.80; or almost perfect, 0.80-1.00.

There was no significant statistical difference between the weights of the patients included in the studies performed with the different techniques including angular modulation (mean weight, 74 kg; range, 44-120 kg), combined modulation (weak decrease [slim]-strong increase [obese] type: mean weight, 75 kg and range, 40-136 kg; average decrease [slim]-average increase [obese] type: mean weight, 78 kg and range, 34-127 kg), or constant tube current (mean weight, 82 kg; range, 44-117 kg) (p = 0.06-0.3). Likewise, no significant difference among the ages of the patients included in the studies performed with the different techniques was found (p = 0.2-0.5).

The average values and SDs for subjective image noise, diagnostic acceptability, and conspicuity of small structures for studies performed with the different techniques are summarized in Table 1. There was no significant difference in image quality parameters (subjective noise, diagnostic acceptability, and conspicuity of small structures) for the examinations performed with constant tube current, angular modulation, and the weak decrease-strong increase type of combined modulation technique (p = 0.01-0.9, Bonferroni's correction) (Figs. 2A, 2B, 2C, and 2D). However, subjective image quality scores for studies performed with the average decrease (slim)-average increase (obese) type of combined modulation were significantly lower than the corresponding scores for the studies performed with the other techniques (p < 0.0001). Among examinations performed with the average decrease (slim)-average increase (obese) type of combined modulation technique, six of nine examinations with more than acceptable subjective noise were not fully acceptable to both radiologists.

View Larger Version

TABLE 1: Average Subjective Image Noise, Diagnostic Acceptability and Conspicuity of Small Structures for Examinations Performed Using Different Techniques

There was no significant difference between patient weight for examinations with unacceptable noise and patient weight for examinations with acceptable noise (p = 0.9). All examinations performed with the techniques, including the weak decrease (slim)-strong increase (obese) type of combined modulation, were deemed fully acceptable.

There was no significant difference in subjective image noise, diagnostic acceptability, and conspicuity of small structures at different levels in the abdomen and pelvis assessed in the present study for examinations performed with different techniques (p > 0.1). However, regardless of the technique used for CT, significantly greater noise streaking artifacts were noted at the level of the confluence of the hepatic veins compared with the porta hepatis and upper margin of the acetabulum (p < 0.0001). There was almost perfect interobserver agreement between the two radiologists (κ =0.9, p < 0.05).

The objective image noise values (SD of the mean attenuation value) for the examinations performed with the different techniques are summarized in Table 2. There was no significant difference in objective image noise values for examinations performed with the constant tube current technique, angular modulation, and weak decrease (slim)-strong increase (obese) type of combined modulation (p > 0.1). However, objective image noise was significantly higher in examinations performed with the average decrease (slim)-average increase (obese) type of combined modulation technique when it was compared with the other techniques assessed in the present study.

View Larger Version

TABLE 2: Average Objective Image Noise, CT Dose Index Volume (CTDI_vol) and Dose–Length Product (DLP) for Examinations Performed Using Different Techniques

The average and SD values of CTDI_vol and DLP for CT examinations performed with different tube current techniques assessed in the present study are summarized in Table 2. A significant reduction in radiation dose was noted with the combined modulation and angular modulation techniques compared with the constant tube current technique (p < 0.0001). Compared with the constant tube current technique, there was a 19% reduction (from 19.0 to 15.4 mGy) in radiation dose for angular modulation, a 42% reduction (from 19.0 to 11.0 mGy) with weak decrease (slim)-strong increase (obese) type of combined modulation, and a 44% reduction (from 19.0 to 10.6 mGy) with average decrease (slim)-average increase (obese) type of combined modulation. The average reduction in radiation dose (p < 0.0001) with combined modulation techniques was 43%, whereas use of the angular modulation technique resulted in a 19% reduction (p < 0.0001).

A significant linear statistical correlation was found between patient weight and CT-DI_vol for studies performed with weak decrease (slim)-strong increase (obese) type (r = 0.84, p < 0.0001) and average decrease (slim)-average increase (obese) type of combined modulation (r = 0.65, p < 0.0001). However, no significant statistical correlation was found between patient weight and CTDI_vol for examinations performed with the constant tube current technique (r = 0.01, p = 0.2) or angular modulation technique (r = 0.19, p > 0.05).