| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
Department of Radiology, University Hospital Basel, Petersgraben 4, Basel, Switzerland.
Received June 27, 2004;
accepted after revision October 22, 2004.
Address correspondence to U. Studler
(studleru{at}tiscali.ch).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Fifty consecutive patients referred for CT of the chest underwent 16-MDCT and, subsequently, axial high-resolution CT. The volumetric raw data from the MDCT scans were reconstructed into slices 2-mm thick using a high-spatial-frequency reconstruction algorithm. Two blinded reviewers independently rated the images from both methods for subjective image-quality criteria. The results were tested for statistical significance using Wilcoxon's signed rank test, and p values of less than 0.05 were considered significant. The effective dose for axial high-resolution CT and volumetric MDCT was calculated.
RESULTS. Motion artifacts were significantly more common on high-resolution CT scans than on reconstructed thin-section images (p < 0.001). The two methods differed significantly in lung attenuation (p = 0.008), mainly because of the presence of ground-glass opacities. The assessment of ground-glass attenuation was superior on axial high-resolution CT. The effective doses were 3.8 mSv for MDCT and 0.9 mSv for high-resolution CT.
CONCLUSION. Reconstructed high-resolution images generated from a single MDCT data acquisition are of comparable quality to images obtained using conventional axial high-resolution CT. However, contiguous MDCT is not recommended for diseases showing predominantly ground-glass patterns, because axial high-resolution CT delineates ground-glass attenuation significantly better.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The advantage of the newly introduced 4 (or more)-MDCT over single-detector CT consists in a high-volume coverage speed without a loss of diagnostic image quality. This advantage has been shown for different section thicknesses, including high-resolution CT [7, 8]. A reduction of threefold or even more in scanning time of the chest means, for the patient, a bearable breath-hold duration and, for the radiologist, fewer breathing artifacts. Furthermore, the same raw data set obtained from MDCT using thin sections can be reconstructed to both volumetric thick-section images and thin-section highly resolved images using a high-spatial-frequency reconstruction algorithm. This technical advantage of MDCT may allow delivery of all required information about the chest in a single CT scan, with the further benefit of sparing the patient the radiation dose of an additional high-resolution CT examination. The purpose of this study was to assess whether quality differed significantly between images obtained using reconstructed volumetric data and images obtained using axial high-resolution CT.
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The patients underwent chest CT for the following known or suspected abnormalities: idiopathic pulmonary fibrosis (n = 4), sarcoidosis (n = 5), bronchiolitis obliterans (n = 4), lymphangitic carcinomatosis (n = 5), tuberculosis (n = 3), lung involvement in collagen vascular disease (n =3), various pulmonary complications after stem cell transplantation (n = 9), bronchiectasis (n =6), emphysema (n = 8), chronic eosinophilic pneumonia (n = 1), idiopathic pulmonary hemorrhage (n = 1), and cryptogenic organizing pneumonia (n = 1). None of the patients with known interstitial lung disease had been evaluated by simultaneous high-resolution CT and MDCT solely for the purpose of assessing disease activity or of monitoring treatment. All patients presented with unexplained symptoms such as acute dyspnea (n = 15), fever (n = 5), and weight loss (n = 3); thus, when the underlying interstitial lung disease had been taken into account, the specific question that led to the use of both high-resolution CT and MDCT in these patients was whether pulmonary embolism, pneumonia, or lung cancer was present.
MDCT Protocols
All scans were obtained using 16-MDCT (Somatom Sensation, Siemens Medical
Systems) with the patient supine. First, MDCT covering the whole lung was
performed in a caudal-to-cranial direction during one breath-hold, resulting
in a volumetric data set. Second, the same patient underwent noncontiguous
axial high-resolution MDCT at 1-cm intervals. The settings used to obtain the
volumetric data were 120 kV, 125-150 mAs (CARE Dose, Siemens), a section
collimation of 16 x 1.5 mm, a table feed of 30 mm per rotation, and a
rotation time of 0.5 sec per 360° tube rotation, resulting in a beam pitch
of 1.25 according to the definition of Silverman et al.
[9]. From the raw data, thick
contiguous sections of 3-mm collimation were reconstructed using a soft-tissue
kernel and a lung kernel for routine diagnostic practice. In addition, a
series of noncontiguous images consisting of the thinnest possible
collimation, 2 mm, was reconstructed at 10-mm intervals using a
high-spatial-frequency reconstruction algorithm (B 80s) for this study. For
axial high-resolution CT, a collimation of 2 x 1 mm was used. A
collimation of 1 x 1 mm was not possible because of the geometric
configuration of the detector array on this scanner. The other settings for
axial high-resolution CT were 120 kV, 120 mAs, and a rotation time of 0.75
sec. Images were reconstructed with a high-spatial-frequency algorithm (B
80f). Axial high-resolution CT was not performed during a single breath-hold
for each image. Instead, depending on the patient's ability to hold the
breath, we scanned during two or three breath-holds, between which the patient
was allowed to breathe. The scanning time for both methods was recorded by the
scanner.
Image Analysis
All images were transferred to a Magic View 1000 workstation (Siemens) for
Windows (Microsoft) and were displayed on the monitor for interpretation. The
scans were displayed at lung window settings (level, -500 H; width, 1,400 H)
and at soft-tissue window settings (level, 40 H; width, 350 H) for evaluation
of the pleura. Images were analyzed independently by two experienced,
board-certified radiologists. The two reviewers were un-aware of the clinical
information, patient data, and acquisition technique. To avoid recall biasing,
the interpretation sessions for high-resolution CT and MDCT images were
separated by at least 4 weeks.
|
In a next step, the reviewers looked for the following abnormal conditions: reticular patterns, small nodules, areas of decreased lung attenuation (bulla, emphysema, or cysts), areas of increased lung attenuation (ground-glass attenuation, consolidation, or mosaic perfusion), and bronchiectasis. Both anatomic and pathologic conditions were rated by a scale ranging from 1 (best) to 5 (worst). In addition, a score of 0 was used for absence of abnormal findings. Subsequently, the images were evaluated for evidence of artifacts, including streak artifacts and respiratory or cardiac motion artifacts (i.e., doubling of fissures, vessels, or bronchi and blurring of the contour of the heart, mediastinum, or diaphragm). The motion artifacts were not graded using a scale. Instead, the reviewers were told to record the number of nondiagnostic images (Figs. 1A, and 1B).
Finally, the two reviewers were asked to assess the overall image quality of the entire study using the same 5-point scale. In summary, using 11 categories for each scan, both reviewers had to rate the image quality of 50 scans obtained by each method.
Radiation Dose
To compare the radiation exposure, the effective doses for the volumetric
and axial techniques were calculated using commercially available software
(WinDose, Scanditronix-Wellhoefer) that was derived from Monte Carlo
calculations [10].
Statistical Analysis
Statistical analyses were performed using SPSS software (release 10.1.3,
SPSS, Inc.). The image quality scores from corresponding categories were
summed for each reviewer, resulting in a total score for each CT technique.
Wilcoxon's signed rank test was used to test pairwise for significant
differences between these total scores, and p values of less than
0.05 were considered to indicate a statistically significant difference.
Weighted kappa was used to assess the rating agreement between the two
reviewers for corresponding categories of each method.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The mean effective dose was 3.8 mSv (3.6 mSv for men and 3.9 mSv for women) for volumetric MDCT and 0.9 mSv (for both men and women) for the axial method.
Image and Statistical Analyses
The histograms in Figure 2
show the distribution of scores for each reviewer and method for 10
categories. The category "artifact" was ignored because this data
type was not ordinal and could take on a continuous range of values. The shape
of the histograms shows a skewed distribution of scores.
Pairwise comparison of scores for normal lung structures showed no significant difference between the two methods. A good image quality for these four categories was expressed by a median score of 2.0 for both axial high-resolution CT and volumetric MDCT. The scores for the category "increased lung attenuation" were significantly higher for reconstructed images from volumetric MDCT than for the corresponding axial high-resolution CT scans (p = 0.0082). This difference in image quality was mainly due to blurring from ground-glass attenuation on the reconstructed images of some patients (Figs. 3A, and 3B). Among the other assessed alterations of lung parenchyma, including reticular changes, small nodules, increased lung attenuation, and bronchiectasis, no significant differences were found (Figs. 4A, and 4B). Overall image quality was rated slightly better for volumetric MDCT than for axial high-resolution CT. However, this difference was not statistically significant. The comparison of the number of motion artifacts achieved statistical significance (p= 0.0004). Both reviewers noted more artifacts on images acquired using axial high-resolution CT (Figs. 1A, and 1B).
|
|
|
|
The average interobserver agreement was 0.6 ± 0.16 (range, 0.39-0.82) for axial high-resolution CT and 0.59 ± 0.15 (range, 0.36-0.80) for the volumetric method.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
We performed the examinations with a CT unit from the latest generation of MDCT scanners, consisting of 16 detectors with an adaptive array configuration. The use of a 1.5-mm collimation in combination with a pitch of 1.25 resulted in a mean scanning time of 6 sec for the entire thorax, a duration for which most patients—even those severely dyspneic—could hold the breath. This contribution to the patient's comfort is reflected in a statistically significant reduction of motion artifacts on images obtained using MDCT, compared with images obtained using axial high-resolution CT (Figs. 1A, and 1B). We did not classify the artifacts by their various origins; instead, we considered the number of artifacts reducing image quality. This approach seems reasonable because prevention of cardiac motion artifacts using ECG-triggered thin-section CT has been shown not to improve diagnostic accuracy [14, 15]. Hence, the diagnostic relevance of heart movement artifacts might be questioned.
Our findings contrast with the previous observations of Kelly and colleagues [13] that volumetric MDCT causes more motion artifacts than does axial high-resolution CT. Most of the patients in that study were scanned with 4-MDCT using a gantry rotation time of 0.8 sec and a slice width of 1.25 mm. The equipment used in that study had to result in a scanning time of up to 20 sec—compared with 6 sec in this study—to cover the chest, although no information about scan duration, pitch, or table feed was provided. Their technique used a single breath-hold for each image obtained with axial high-resolution CT, whereas we acquired high-resolution CT scans using two or three breath-holds with respiration allowed in between, resulting in a longer breath-hold period (average breath-hold time, 9.7 sec, compared with 6 sec using MDCT) and, consequently, in more motion artifacts. Using our high-resolution CT protocol, the gantry rotation time was 0.25 sec longer (0.75 sec) than the time needed for a 360° tube rotation using the MDCT protocol (0.5 sec). However, this slight time difference has little effect on substantial artifacts; if it has any effect, that effect may be on cardiac motion artifacts. In contrast to Kelly et al., we applied a 1-cm spacing interval, resulting in a doubling of the number of images to be interpreted and a more pronounced effect on the statistical analysis. Altogether, the longer scanning time (by three to four times) used for 4-MDCT in the work of Kelly et al. is the most important factor to consider when evaluating the discrepancy between our results and theirs.
Schoepf et al. [6] compared the quality of thin-section images obtained from a single-detector scanner to that of reconstructed images obtained from an MDCT scanner and found no significant difference. However, the MDCT scanner they used required a mean duration of 27 sec to cover the entire lung, a situation hardly comparable to that of our study regarding artifacts. Furthermore, because the two techniques were compared in two different patient populations, those results could have been biased. Our study setting permitted both methods to be performed on the same patients, excluding this kind of bias.
A significant difference in increased lung attenuation between the two methods indicated that high-resolution CT better depicted the presence of ground-glass attenuation (Figs. 3A, and 3B). We performed MDCT with a collimation of 1.5 mm, for which the thinnest possible image reconstruction is 2 mm. Another option would have been to use the thinner, 0.75-mm collimation with the same pitch. This protocol offers the advantage of 1-mm-thick slice reconstructions and, therefore, may reduce blurring and improve image detail without a significant increase in radiation dose. However, the use of a narrower collimation halves the table speed and increases the tube rotation time, resulting in longer scans and, consequently, more artifacts. In addition, the huge amount of data generated by such protocols becomes increasingly difficult to manage. We believe that the parameters used, including a collimation of 1.5 mm, were appropriate for the imaging quality required, and although image resolution with this protocol was significantly reduced because of ground-glass opacities, neither reviewer considered the overall image quality restricted.
The average interobserver variability as measured by weighted kappa was moderate to good for both methods. The two lowest values for either method were in the category "artifacts." The values were less than 0.4, indicating poor agreement. This was attributed to the fact that one reviewer used more rigorous criteria to characterize images containing motion artifacts as nondiagnostic than the other reviewer did. However, kappa values from both methods exceeded chance levels.
The calculated effective doses—3.8 and 0.9 mSv for volumetric MDCT and axial high-resolution CT, respectively—were similar to those found in other studies [6, 13]. The higher effective dose received from axial high-resolution CT in this study than in the study of Kelley et al. [13] (0.9 vs 0.58 mSv) was due to the different spacing intervals in our study and theirs (10 and 20 mm, respectively).
A potential limitation of our study was that it was not designed to assess the diagnostic accuracy of images obtained from the two methods. Although all the important high-resolution CT criteria for normal and pathologic lung parenchyma structures were included in the rating, our results are based on subjective image interpretation.
An additional limitation of the study involves the different collimations and reconstruction thicknesses used between the two methods. The MDCT collimation of 1.5 mm and the resulting reconstruction slice thickness of 2 mm constitute a compromise between a short examination time and the slice width. The different slice thicknesses soon may be fully compensated for when the faster 32- or 64-MDCT scanners are implemented into routine clinical practice.
In summary, the current study has revealed that reconstructed thin-section images of good quality can be obtained using a standard chest MDCT protocol. However, reconstructed thin sections obtained from MDCT cannot be recommended for diseases showing predominantly ground-glass patterns.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  A. A. Bankier, S. Mehrain, D. Kienzl, M. Weber, M. Estenne, and P. A. Gevenois Regional Heterogeneity of Air Trapping at Expiratory Thin-Section CT of Patients with Bronchiolitis: Potential Implications for Dose Reduction and CT Protocol Planning Radiology, June 1, 2008; 247(3): 862 - 870. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Madani, V. De Maertelaer, J. Zanen, and P. A. Gevenois Pulmonary Emphysema: Radiation Dose and Section Thickness at Multidetector CT Quantification--Comparison with Macroscopic and Microscopic Morphometry Radiology, April 1, 2007; 243(1): 250 - 257. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. A. Bankier, C. Schaefer-Prokop, V. De Maertelaer, D. Tack, P. Jaksch, W. Klepetko, and P. A. Gevenois Air Trapping: Comparison of Standard-Dose and Simulated Low-Dose Thin-Section CT Techniques Radiology, March 1, 2007; 242(3): 898 - 906. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
