| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Both authors: Department of Radiology, Hitachi General Hospital, Jyonanncho 2-1-1, Hitachi City, Ibaraki 317-0077, Japan.
Received May 31, 2004;
accepted after revision September 15, 2004.
Address correspondence to T. Irie
(toshiyuki.irie{at}ibabyo.hitachi.co.jp).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Contrast-enhanced abdominal CT images obtained by a 4-MDCT scanner were retrospectively analyzed. The image-acquisition factors were fixed (group A, n = 104). The image noise was measured at five regions: the aorta, the back muscle, the spleen, and the peripheral and the central hepatic portions. Body size was evaluated by two methods on the basis of body weight (method A) and on the scout image data (method B). Coefficients of determination of 10 relationships between body size and image noise were calculated. For the next study, CT images were prospectively obtained by modulating the mAs on the basis of body weight (group B, n = 100) and the scout image data (group C, n = 100). We compared the differences in SDs of aortic image noise in each group using the F test.
RESULTS. The coefficients of determination of the aorta were 0.619
and 0.812 for methods A and B, respectively, and the highest values in both
methods (p = 
2, Fisher's exact test). The SDs of
the aortic image noise of groups A, B, and C were 2.03, 1.61, and 1.06,
respectively. There were statistically significant differences between each
group (p < 0.022), and the SD was the smallest in group C.
CONCLUSION. The aorta was the optimal region to measure the image noise. Individual mAs modulation based on the scout image data was useful to obtain images of similar noise levels.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our purposes were to determine the most reliable region to measure the image noise, to evaluate the patient's body size for modulation of the mAs, and to confirm the efficacy of the individual mAs modulation to obtain images of similar noise levels.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
 |
The second method was to evaluate the cross-sectional dimension and
anteroposterior and transverse body diameters using the scout image data for
the calculation of the mean body diameter. Every pixel in the scout image data
included the value of the X-ray attenuation from a focal spot to a detector
element. As for the CT machine in this study (Robusto, Hitachi Medico), the
value of a pixel (P) of the scout image could be measured and given
as:
 |
Practically, an elliptic ROI was put at the level of the liver hilus, and
the value was measured (Fig.
1). The horizontal diameter of the ROI was approximately half of
the transverse body diameter and relatively large to minimize the influence of
bone density on the scout value. The transverse body diameter on the scout
image (Xs) was also measured at the same level of the elliptic ROI,
and the actual transverse body diameter (X) was given as X =
Xs / M, where M was a magnification factor on the
scout image. Finally, D was given as:
 |
|
Analysis of the Relationship Between CT Image Noise and Mean Body Diameter
CT image noise was given as:
 |
 |
And for method B, the equation was:
 |
Linear regression could be applied to the analysis of the relationship
between the image noise and the body size in each method: [D =
(S x Xs)
]. When the mAs value was
known or fixed, this analysis enabled calculation of the values of C,
Ac, and Ac x B. After these values were
calculated and the target level of image noise was defined, the optimized mAs
could be calculated for the prospective modulation on the basis of the body
size: [D = (S x Xs)].
Determination of the Optimal Region to Measure Image Noise
Clinical CT images obtained between July and August 2003 were
retrospectively analyzed. Patient ages ranged from 16 to 85 years (mean, 61.8
years) and body weight, from 37 to 86 kg (mean, 56. kg); 48 females and 56
males were included in this analysis. Patients with liver cirrhosis, fatty
liver, or ascites were excluded from this study. The image-acquisition factors
were the following: 4 x 2.5 mm collimation, 0.8 sec/rotation, 17.5
mm/rotation table movement, 120 kV/250 mA tube voltage/mA (200 mAs). The image
reconstruction factors were 5 mm in thickness with use of a standard
reconstruction kernel. Prefilled syringes of contrast medium ([100 or 150 mL
of iohexol] Omnipaque 300, Daiichi Seiyaku) were used. The amount of the
contrast medium was 100 mL when body weight was less than 60 kg; 2 mL/kg, when
60 kg or more but less than 75 kg; and 150 mL, when more than 75 kg. The
contrast medium was injected via a plastic needle placed in the antecubital
vein at a rate of 2 mL/sec, and scanning was started 25 sec after the end of
injection to obtain portal venous phase images
[13,
14].
We analyzed the CT images at the level of the liver hilus. Image noise was measured at five regions: the aorta, peripheral and central portions of the liver, the spleen, and the back muscle. For the measurement of the aorta, the ROI was placed in the center of the aorta, and the diameter was approximately 75% of the aortic diameter. For the measurement of the liver and the spleen, the ROI was placed to avoid the vessels and streak artifacts from the ribs, and the diameter was 1–1.5 cm. For the measurement of the back muscle, the shape of the ROI was often elliptic to avoid fat among the muscle bundles. When the spleen was removed (two cases) or the back muscle included much fat tissue (four cases), only the data of other organs were included in the analysis.
Body size was evaluated by the two methods, and a total of 10 relationships between image noise and body size were examined by linear regression. Each coefficient of determination was calculated.
Prospective Modulation of the mAs
This section of the study was approved by the ethics committee of our
hospital. The mAs value was individually modulated on the basis of body size
to achieve similar levels of image noise. The mean value of the image noise in
the previous section of the study was used as the target level. The data of
the previous section of the study were also used for comparison as a fixed mAs
group (group A, n = 104). We prepared two charts: one for modulation
based on body weight and another, on the scout image data. The gantry rotation
speed was 0.8 sec when the mAs was 240 or less and 1.5 sec when the mAs was
300 or more. The other image acquisition factors were identical to those in
the previous section of the study.
Between September and October 2003, the mAs was modulated on the basis of body weight (group B, n = 100; 44 females and 56 males). Patient ages ranged from 15 to 89 years (mean, 63.9 years) and the body weight ranged from 3 to 80 kg (mean, 54.0 kg). Two subjects were excluded from the study because body weight was more than 86 kg and the calculated mAs exceeded the maximum output (450 mAs) of the CT machine.
Between October and November 2003, the mAs was modulated on the basis of the scout image data (group C, n = 100; 43 women and 57 men). Patient ages ranged from 20 to 85 years (mean, 61.9 years) and the body weight ranged from 34 to 76 kg (mean, 55.2 kg). Seven cases were excluded from the study because the calculated mAs exceeded the maximum output of the CT machine.
The difference in the SD of the image noise was compared by the F test between each group. The differences in the image noise among the groups were compared by the nonparameteric Kruskal-Wallis test. The threshold of the p value was 0.05.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2, Fisher's exact
test).
|
|
|
Prospective Modulation of the mAs
The mean image noise of the aorta in the previous section of the study was
10.95. This level was used as the target level in this section. Two charts,
one for mAs modulation based on body weight
(Table 2) and another based on
the scout image data (Table 3),
were prepared for this section of the study. The SDs of the aortic image noise
of groups A, B, and C were 2.03, 1.61, and 1.06, respectively (Fig.
3A,
3B,
3C). There were statistically
significant differences in the SD between group C and groups A and B
(p < 0.0001, F test) and between groups A and B
(p = 0.0218, F test). The SD of the aortic image noise was
minimized when the mAs was modulated on the basis of the scout image data. The
mean image noise levels of groups A, B, and C were 10.95, 10.82, and 10.56,
respectively. There were no statistically significant differences among the
groups (p = 0.47, Kruskal-Wallis test).
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Discovery of the optimal region to measure the CT image noise enabled quantitative analysis of clinical CT images. As an example shown in this study, we could develop an mAs chart for the image acquisition of similar noise levels. This chart for the mAs modulation based on the scout image data is now used in practice in our hospital (Table 3). This chart could also be used to perform reduced-dose CT. There were many reports to show the same or similar diagnostic efficacy of the reduced-dose CT compared with the standard-dose one [15–19]. However, a concern is that the CT images with a reduced dose might be too noisy in large-sized patients. Using the mAs chart, the mAs could be decreased with confidence that the image would not be too noisy. For example, in our hospital the mAs is decreased by 50% in cases of repeated follow-up of patients with acute pancreatitis or acute aortic dissection. The mAs is also decreased by 50% in conventional CT examinations when a subsequent contrast-enhanced CT examination is planned.
How much is the optimized value of the aortic SD? We still have no answer
based on scientific research. However, the degree of the image mottle
correlated well with the aortic SD, and approximate visual inspection and
impression were as follows: unacceptable (aortic SD was > 14.5), minimally
acceptable (13–14.5), appropriate (9–13), and excellent (
9).
The 100 patients' images obtained by individualized mAs modulation based on
the scout image data were categorized as appropriate (94%) or excellent (6%).
Although excellent images were acceptable for diagnosis, the radiologists
should pay attention not only to the poor quality images but also to the
excellent images. At the expense of the excellent images, the patients might
receive too much radiation. For scientific research to decide the optimized
value of the aortic SD, we are interested in the image creation by
computer-simulated radiation-dose reduction
[20]. By incrementally adding
noise to the clinical images with pathologic findings, the maximized aortic SD
and the minimized radiation dose for lesion detection could be determined.
As shown in this study, mAs modulation based on the scout image data was superior to that based on body weight in achieving consistent levels of image noise. In previous studies, body weight was used [7, 10] or actual transverse diameter was measured using a slide caliper to modulate the mAs [9]. In another previous study, the concept of the scout density was used to calculate the mean diameter of the body section [6]. This method was sophisticated but seemed impractical because a separate workstation was required for calculation. Compared with these previous methods, our method was practical and needed no additional equipment. Each original mAs chart could be developed for each institution to negotiate the differences in CT machines and image acquisition protocol by following our method and using the routine clinical images. One problem is that the software to measure the distance and the density on the scout image might not be available in other CT machines. To solve this problem, we believe a preparatory single CT slice acquired with minimized mAs might be useful to evaluate the mean body diameter for the mAs modulation. The chart for the mAs modulation based on the axial CT image could be made by referring to our method.
In our study, mAs modulation was performed for a single-slice level. However, the cross-sectional dimensions of the body differ at each slice level. Some modern CT scanners are equipped with automatic mAs modulation technology to render all images with similar noise levels, independent of the patient's size and anatomy [3, 21]. It is desirable that all CT scanners use similar automatic modulation technology in the future.
It is important to estimate and quantify the radiation dose to the patient. For this purpose, the dose-length product (DLP) is a convenient quantity. The effective dose can be estimated by multiplying the appropriate conversion factor to the DLP [22]. However, there is a limitation in the DLP. The value of DLP is calculated on the basis of the CT dose index (CTDI), and the current American Association of Physicists in Medicine protocol for the measurement of CTDI requires the use of only two different sizes of the acrylic phantoms: a 16-cm cylindric phantom simulating an adult's head and a 32-cm cylindric phantom simulating an adult's body. When the CT acquisition factors (mAs, beam collimation, and total acquisition time) are the same, the body diameter of the patients does not influence the value of DLP or CTDI. However, phantom studies showed that the mean imparted section dose increased when the diameter was smaller [11, 23]. Thus, the small pediatric patient receives more radiation dose compared with the adult patient when the CT machine displays same value of CTDI or DLP.
In conclusion, the aorta was the most reliable region for measuring the CT image noise in contrast-enhanced abdominal CT. Evaluation of both the transverse diameter and the scout density was useful for individual mAs modulation to obtain similar levels of image noise. It is still unclear how much image noise is ideal to minimize the radiation dose while maximizing the diagnostic efficacy. Further clinical studies will be necessary to establish the optimized level of image noise.
Acknowledgments
We thank Hiroshi Takagi and Yasushi Miyazaki (Hitachi Medico, Kashiwa) for
the information about CT.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Herzog, D. M. Mulvihill, S. A. Nguyen, G. Savino, B. Schmidt, P. Costello, T. J. Vogl, and U. J. Schoepf Pediatric Cardiovascular CT Angiography: Radiation Dose Reduction Using Automatic Anatomic Tube Current Modulation Am. J. Roentgenol., May 1, 2008; 190(5): 1232 - 1240. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Hur, S. W. Hong, S. Y. Kim, Y. H. Kim, Y. J. Hwang, W. R. Lee, and S. J. Cha Uniform Image Quality Achieved by Tube Current Modulation Using SD of Attenuation in Coronary CT Angiography Am. J. Roentgenol., July 1, 2007; 189(1): 188 - 196. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
