| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 All authors: Department of Diagnostic Radiology, Yale University School of Medicine, 333 Cedar St., PO Box 208042, New Haven, CT 06520-8042.
Received August 12, 2007;
accepted after revision January 28, 2008.
Address correspondence to G. M. Israel
(gary.israel{at}yale.edu).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We evaluated the CT scans of patients who underwent imaging of the abdomen and pelvis using dose modulation and the fastest gantry rotation time available on two different CT scanners. Ninety-five patients were scanned with a 64-MDCT scanner (noise index, 11; tube rotation speed, 0.5 second) with a maximal x-ray tube capacity of 695 mA. Ninety-four patients were scanned with a 16-MDCT scanner (noise index, 11.6; tube rotation speed, 0.6 second), which has a maximal x-ray tube capacity of 440 mA. The total number of images per examination, total number of images obtained at saturated tube current, image noise (SD of fluid attenuation), and patient width were recorded. A qualitative evaluation of image quality, with images obtained below and at the maximal tube current grouped separately, was performed by two independent radiologists who were not blinded to the type of scanner used using a scale of from 1 (best) to 4 (worst). Statistical analyses included the Kruskal-Wallis one-way analysis of ranks test for nonparametric ordinal data, the unpaired two-tailed Student's t test, and the chi-square test.
RESULTS. For images obtained with the stronger x-ray tube (maximum tube current = 695 mA), the average number of axial images per examination was 87.6. In 34 of 95 (36%) patients, at least one image was acquired with the tube current saturated. The average image noise was 12.4 H. Subjective evaluation yielded an average image quality score of 1.2 for images below saturated tube current and 1.2 for images at saturated tube current. For images obtained with the weaker x-ray tube (maximum tube current = 440 mA), the average number of axial images per examination was 88.9. In 84 of 94 (89%) patients, at least one image was acquired with the tube current saturated. The average image noise was 16.8 H. Qualitative evaluation showed average image quality scores of 1.3 and 1.8 for images below and at the saturated tube current, respectively. The percentage of images acquired at the saturated tube current was significantly greater for the weaker x-ray tube than the stronger x-ray tube (p < 0.0001), and qualitative analysis of images obtained at saturated tube current showed significantly decreased quality for the weaker x-ray tube when compared with images obtained with nonsaturated current (p = 0.001).
CONCLUSION. On the MDCT scanners investigated, when dose modulation is combined with fast tube rotation times, tube current saturation occurs with weaker x-ray tubes resulting in deterioration of image quality.
Keywords: abdominal imaging • CT technique • dose modulation • gantry rotation time • MDCT • tube current
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Today, CT scanners are capable of subsecond gantry rotation times. When fast gantry rotation times are combined with tube current modulation, it is possible to saturate CT tube current (tube current reaching its maximal capacity) while the scanner tries to maintain constant image noise. Tube current saturation results in underexposing patients with subsequently increased image noise. In some cases of suspected renal colic or in CT colonoscopy, the increased noise may not have a deleterious effect on image quality [7, 8]. However, in other cases, the increased image noise may result in suboptimal image quality, which can adversely affect patient care [9, 10]. The purpose of this study was to determine, using one manufacturer's 16- and 64-MDCT scanners, the degree to which CT tube current saturates if tube current modulation and fast gantry rotation speeds are used when imaging the abdomen and pelvis and to determine if saturated tube current affects image quality.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
18 years) per scanner in which a routine
CT examination of the abdomen and pelvis was performed. Each examination was
retrieved on a PACS (Synapse, Fuji Medical Systems), and only patients imaged
at the fastest gantry rotation time (64 detector, 0.5 second; 16 detector, 0.6
second) were included (64 detector, 95 patients; 16 detector, 94 patients) in
the study. The following information was obtained from each CT examination and recorded on a data sheet by a single author: total number of axial images acquired, total number of axial images acquired with the tube current saturated (tube current reaching its maximal mA setting as defined later), the Hounsfield unit attenuation in the gallbladder (n = 41, 64 detector; n = 15, 16 detector) or urinary bladder (n = 54, 64 detector; n = 79, 16 detector) and its SD (image noise), and the largest width of the patient at the level of the liver. For the attenuation of fluid in the gallbladder or urinary bladder, the gallbladder was the initial choice for this measurement. If the patient had a previous cholecystectomy, the urinary bladder was used for the measurement. For the measurement of image noise (SD of fluid in the gallbladder or urinary bladder) on the 64-MDCT scanner, 72 measurements were made on nonsaturated images and 23 on saturated images. For the 16-MDCT scanner, 14 measurements were made on nonsaturated images and 80 on saturated images.
For each scan, two radiologists, one with 11 years of body CT experience and the other with 2 years of body CT experience, in consensus performed a qualitative evaluation of image quality among all images obtained below and at tube current saturation using a 4-point scale: 1, outstanding image quality without significant noise seen on the images; 2, good image quality with noise mildly limiting interpretation of images; 3, fair image quality with noise moderately limiting interpretation of images; and 4, poor image quality with noise severely limiting inter pretation of images. For this assessment, all of the images obtained below and at tube current saturation for each patient were evaluated separately and a single rating was given for each group. If a group of images (either below or at tube current saturation) contained images that were of good and poor quality, the radiologists gave a single subjective grade, taking into account both the good- and poor-quality images. The radio logists were not blinded to the type of scanner used, and the examinations from the two different scanners were not intermixed.
The LightSpeed VCT (64 detector) is equipped with an x-ray tube (Performix Pro 100, GE Healthcare) capable of a maximal tube current of 800 mA; this scanner will be referred to as the "high-tube-current scanner." The maximal current of the tube was limited to 695 mA, as recommended by the vendor, in an effort to limit radiation dose. All examinations performed on the VCT were obtained using tube current modulation in the x-, y-, and z-axes (SmartmA, GE Healthcare) with a noise index of 11 (suggested by the manufacturer) and the following parameters: slice thickness, 5 mm; slice interval, 5 mm; table speed per rotation, 40 mm (pitch = 1); and 120 kVp.
The LightSpeed CT (16 detector) is equipped with a tube (Performix, GE Healthcare) capable of a maximal tube current of 440 mA; this scanner will be referred to as the "low-tube-current scanner." All examinations performed on the Light-Speed CT were obtained using tube current modulation in the z-axis (AutomA, GE Healthcare) (x-, y-axis modulation was not available) with a noise index of 11.57 (suggested by the manu facturer) and the following parameters: slice thickness, 5 mm; slice interval, 5 mm; table speed per rotation, 27.5 mm (pitch = 1.375); and 120 kVp.
The noise index is a manufacturer-specific parameter that allows a radiologist to select the amount of x-ray noise that will be present in the reconstructed images. The noise index is approximately equal to the SD of water in the central region of the image when a uniform phantom is scanned and reconstructed using the standard reconstruction algorithm. As the noise index increases, the required mA decreases and image noise increases.
Statistical Analysis
Using the Kruskal-Wallis one-way analysis of ranks test for nonparametric
ordinal data, a comparison of qualitative score assigned with regard to image
quality of images obtained at saturated and nonsaturated tube current was
performed for the data obtained from each scanner. An unpaired two-tailed
Student's t test was used to compare the quantitative measurement of
image noise of images obtained at saturated and nonsaturated tube current from
each scanner and to compare the measurement of patient widths from each
scanner. The chi-square test was used to compare the percentage of images
obtained at a saturated tube current on each scanner and the percentage of
examinations performed on each scanner with at least one image obtained at a
saturated tube current. Linear correlation was performed to assess the
relationship between the number of slices in an examination and the percentage
of slices scanned at the maximum tube current. Separate analysis was performed
for each scanner. Pearson's correlation coefficient was computed, and
significance was assessed via a two-tailed p value. Bonferroni
correction for multiple comparisons was used so that statistical significance
was accepted for a p value of < 0.0125 (0.05/4).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
For the high-tube-current scanner, the average number of axial images per examination was 87.6 (range, 72–106 images). In 34 of 95 (36%; 95% CI = 26–46%) patients, at least one image (mean, 19.4 images; range, 5–106 images) was acquired with the tube current saturated (Fig. 1). The average image noise was 12.4 H (range, 8–21 H). Subjective evaluation of image quality yielded an average score of 1.2 (range, 1–2) for both image groups (below and at tube current saturation). The average width of the patients was 33.7 cm (range, 27–49.1 cm).
|
|
|
|
|
Quantitative Evaluation of Image Noise
On average, image noise was higher for the images obtained at saturated
tube current than images obtained at nonsaturated tube current on both
scanners. However, this difference was not statistically significant for
either the high-tube-current scanner (p = 0.05) or the
low-tube-current scanner (p = 0.236)
(Table 2).
|
Likelihood of Scanning at Saturated Tube Current
A statistically significant larger percentage of total images on the
low-tube-current scanner (81% [6,696/8,268], 95% CI = 80–82%) was
acquired with tube current saturated than on the high-tube-current scanner
(24% [1,943/8,240], 95% CI = 23–25%) (p < 0.0001) (Figs.
1 and
2). In addition, the number of
cases with at least one image acquired with tube current saturated was
significantly greater for the low-tube-current scanner (89% [84/94], 95% CI =
80–94%) than the high-tube-current scanner (36% [34/95], 95% CI =
26–46%) (p < 0.0001).
Relationship of Number of Slices and Tube Saturation
For the high-tube-current scanner, Pearson's correlation coefficient for
the relationship between the number of slices in the examination and the
percentage of slices scanned at maximum tube current was 0.45 with p
< 0.0001. If Pearson's correlation coefficient is recomputed for the same
data set with the scans having no slices at maximum tube current removed, it
is 0.54 with p = 0.001.
For the low-tube-current scanner, Pearson's correlation coefficient for the relationship between the number of slices in the examination and the percentage of slices scanned at maximum tube current was 0.33 with p < 0.001. If Pearson's correlation coefficient is recomputed for the same data set with the scans having no slices at maximum tube current removed, it is 0.38 with p = 0.0004.
Comparison of Patient Widths from Each Scanner
There was no statistically significant difference of patient widths imaged
on each scanner (p = 0.87).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our study shows that, using one manufacturer's x-ray tubes and MDCT scanners, tube current is saturated when dose modulation is combined with fast rotation speeds during routine CT of the abdomen and pelvis. The degree of tube saturation is dependent on multiple factors including the maximal possible tube current, tube rotation time, desired noise level of the images, and patient size [5, 6]. It should not be surprising that tube saturation occurred more frequently when patients were imaged with an x-ray tube with a maximal 440-mA output (81% of all images) than with an x-ray tube with a maximal 695-mA output (24% of all images). Although the minor differences in tube rotation times and noise indexes used during each examination and variations in patient size could account for part of this, the difference in maximal tube current is likely the major contributor. Superficially, the degree of tube saturation present in this study may appear to depend on the number of detectors that each scanner possesses. However, the major determinate of tube saturation is the maximal output that the tube is capable of.
Tube saturation may be limited by increasing the noise index or increasing the tube rotation time. By increasing the noise index (and keeping the tube rotation time constant), less tube current would be necessary for all images. Theoretically, some of the images initially obtained at the saturated tube current would then be able to be acquired at a nonsaturated level. This change would also decrease overall radiation dose but would result in increased image noise and subsequent decreased image quality. By increasing tube rotation time (and keeping the noise index constant), the temporal resolution of the scanner is decreased and overall radiation dose to the patient will increase because the tube current–time product will increase for images that were initially obtained at tube saturation. This change results in increased radiation dose to the patient but an overall increase in image quality because image noise has decreased for those images no longer obtained at tube saturation. One final method to avoid tube saturation is to increase peak kilo-voltage. This change, however, will increase the effective energy of the x-ray beam and distance it farther away from the maximum absorption of the k-edge of iodine [13].
In our study, images obtained at nonsaturated tube current were of significantly better subjective quality than images obtained with saturated tube current using the low-tube-current scanner (p = 0.001). However, for the high-tube-current scanner, there was no statistical difference for subjective image quality at saturated and nonsaturated tube current. This lack of significant quality difference is likely secondary to the overall higher tube current–time product available with the high-tube-current scanner (695 mA x 0.5 second = 347.5 mAs) as compared with the low-tube-current scanner (440 mA x 0.6 second = 264 mAs). In other words, images obtained at saturated tube current were closer to the necessary tube current–time product (prescribed by the noise index) with the high-tube-current scanner (695-mA maximum) than with the low-tube-current scanner (440-mA maximum).
Curiously, the quantitative analysis of image noise did not show a significant difference between images obtained at saturated versus nonsaturated tube current for either scanner. The reasons for this finding are unclear. Image noise was measured using the SD of fluid (bile or urine), and the SD of x-ray attenuation can be affected by multiple parameters such as the reconstruction algorithm, the x-ray filter, variations in patient anatomy, patient motion, beam-hardening artifacts, detector configuration, detector element size, cone beam angle, and improper centering of the patient in the scanning field of view [5, 6]. Therefore, these variables may have affected the quantitative measurement of image noise. In addition, because the fluid measured (bile or urine) was not a random selection, the SD of the attenuation of bile or urine might be variable.
There are limitations to this study. First, both angular (x- and y-planes) and z-axis modulation were used when imaging with the high-tube-current scanner, whereas only z-axis modulation was available when imaging with the low-tube-current scanner. As opposed to z-axis modulation alone, modulating dose in the x-, y-, and z-planes further decreases overall tube current and radiation dose when imaging the same patient. Unfortunately, it is not possible to determine whether this difference had an effect on our cohort. Also, the noise index, tube rotation times, and pitch were slightly different between the scanners. Although the noise index and tube rotation times favored the low-tube-current scanner, the higher pitch value used with the low-tube-current scanner favors the high-tube-current scanner. Also, during the qualitative analysis, the readers were not blinded to whether the images had been obtained at or below tube saturation, which may have introduced bias. In addition, image noise was quantified by measuring the SD of fluid (gallbladder or bladder), which may show individual and temporal variations in attenuation. There was a high incidence of previous cholecystectomy in the study sample, and for the low-tube-current scanner, image noise was measured in a larger percentage of cases in the bladder. This measurement may be affected by greater radiation absorption of the bony pelvis as compared with the ribs. Also, there may have been variability in the patient size imaged on the different scanners, accounting for some of the differences in tube saturation. However, in our study, there was no statistically significant difference in patient widths between both scanners. Because dose modulation was used in the x-, y-, and z-axes, patient width may not be sufficient for measuring patient size and patient size in the anteroposterior direction may also be important. Also, we did not record whether patients had metallic prostheses, which can affect tube modulation. In addition, image quality evaluation was subjective and interobserver variability may occur. In some cases, an examination may have contained both outstanding and poor-quality images, which may also have contributed to subjective variability. Finally, the results of our study are limited to the two tested scanners by a single manufacturer and might not be generalized to other scanners.
In conclusion, when dose modulation is combined with fast tube rotation times, tube saturation occurs. Tube current saturation results in an increased image noise and subsequent deterioration of subjective image quality. To avoid this decrease in image quality and maintain standard image noise and quality, the prescribed temporal resolution for a CT examination should be considered to ensure the appropriate tube current–time product can be achieved. Use of a slower gantry rotation time may be advantageous for imaging applications that do not demand the highest temporal resolution.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
