| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Department of Radiology, Vancouver General Hospital, University of British Columbia, 899 W 12th Ave, Vancouver, BC, Canada V5Z 1M9 (J.R.M., H.O.C.); Department of Radiology, Pusan National University Hospital, Korea (K.I.K.); Department of Radiology, Christchurch Hospital, New Zealand (S.L.S.M.); Department of Radiology, Osaka University Medical School, Japan (T.J.); Department of Radiology, Wake Forest University School of Medicine, Winston-Salem, NC (P.K.); and Department of Radiology, National Jewish Medical and Research Center, Denver, Colo (S.V.). Received June 6, 2003; revision requested August 20; revision received November 18; accepted January 12, 2004. Address correspondence to J.R.M. (e-mail: jmayo@vanhosp.bc.ca).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: The noise level in the raw scan data of 150 clinically indicated conventional tube current (200–320-mA) chest CT examinations was modified to simulate tube current reduction to 100 and to 40 mA. A total of 450 image sets were thus available. Four radiologists blinded to the tube current used assessed the image sets in random order for 14 structures and lung findings and ranked subjective image quality by using a five-point scale (1 = nondiagnostic, 2 = inferior, 3 = adequate, 4 = good, 5 = excellent). After a 3-week interval, the 150 conventional tube current image sets were rescored so that intraobserver agreement could be assessed. The McNemar statistic was used to determine whether there were more scoring disagreements between interpretations of the conventional and those of the reduced tube current scans or between the two interpretations of the conventional tube current scans.
RESULTS: When overall agreement for 14 structures and lung findings was pooled over four observers, significantly more disagreements (P < .05) were seen when scores were compared between conventional and reduced tube current scans than when scores were compared between repeated interpretations of the conventional tube current scans. There was a significant decrease (P < .05) in the subjective image quality of reduced tube current scans compared with the subjective image quality of conventional tube current scans.
CONCLUSION: These data indicate that reduced tube current does affect reader evaluation of structures and lung findings and reduces a reader’s subjective assessment of image quality.
Supplemental material: radiology.rsnajnls.org/cgi/content/full/2323030899/DC1.
© RSNA, 2004
Index terms: Computed tomography (CT), image quality, 60.12115 • Computed tomography (CT), radiation exposure, 60.12115 • Radiations, exposure to patients and personnel • Thorax, CT, 60.12115
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Designing experiments to determine the relationship between diagnostic accuracy and radiation dose is challenging because additional radiation exposure is undesirable. Although the radiation dose delivered at CT can be modified by changing the tube current (in milliamperes), tube voltage (in peak voltage), scanning time, pitch, or scanner geometry, the most commonly manipulated parameter is the tube current (8). The dose and radiation exposure at CT are linearly related to the tube current if all other parameters are held constant.
Previous investigators have repeatedly scanned selected anatomic levels with conventional and reduced tube currents to compare high- and low-dose CT images. With this experimental design, subjects receive an increased radiation dose because additional reduced tube current scans are obtained. To minimize additional radiation exposure, previous experiments have involved limiting comparisons of different tube currents to three to five selected anatomic levels (9–11). Unfortunately, when scanning the chest, repeated scans obtained at the same table position are often anatomically different because of differences in the level of inspiration (12). This section misregistration may mask the effect of reduced radiation doses. To our knowledge, the comparison of complete sets of perfectly registered conventional and reduced tube current images from helical chest CT examinations to permit a more detailed assessment of the influence of reduced tube current on diagnostic accuracy has not previously been reported.
To address this problem, in conjunction with a major CT equipment manufacturer (GE Medical Systems, Milwaukee, Wis), we developed and validated a computer technique to simulate the increase in image noise that accompanies the use of reduced tube current (12). We note that changes in tube voltage (in kilovolt peaks) and associated changes in beam hardening cannot be simulated by using this computer program. This computer program allows reduced tube current CT scans to be simulated by modifying the raw scan data obtained at higher tube current examinations, with no additional radiation exposure to the subject. When this technique is used, the only difference between the conventional tube current images and the reduced tube current images is the level of random noise in the image. With the exception of the added noise, which increases as the simulated tube current is reduced, section position and any motion artifacts are identical to those on the conventional tube current scans. Thus, the purpose of our study was to assess, by using computer simulation, the effect of reduced tube current on reader evaluation of structures and lung findings on images obtained at clinically indicated chest CT examinations.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The CT examinations were performed because of the following clinical indications: lung cancer (n = 87), metastatic disease (n = 15), mediastinal mass (n = 10), chest infection (n = 7), esophageal cancer (n = 6), hemoptysis (n = 6), chest trauma (n = 5), aortic abnormality (n = 4), chest wall abnormality (n = 4), cardiac abnormality (n = 3) mesothelioma (n = 2), and abdominal pain (n = 1). Enhancement with intravenous contrast material (ioversol, Optiray 320; Mallinckrodt Medical, Pointe Claire, Quebec, Canada) was used in 69 of the 150 examinations. The average number of images reconstructed for review was 42 (range, 28–51).
The tube currents for the diagnostic examination that yielded the raw scan data were varied from 200 to 320 mA by the technologist scanning the patient on the basis of his or her experience with patient size and resultant image quality. This operator-dependent variation in milliamperage was consistent with technologist practice at Vancouver General Hospital at the time of the study. Data from consecutive examinations could not be included because resources available for image processing and storage limited case acquisition to one or two examinations per week.
All raw scan data for the conventional tube current examinations were acquired by using one of two identically equipped commercially available single-section helical CT scanners (Hi-Speed Advantage; GE Medical Systems) with 3.1-level software and a solid-state detector array (of scintillation crystal photodiode). Mediastinal and lung images were reconstructed from the conventional tube current raw scan data by using standard and high-spatial-frequency (bone) algorithms, respectively, and a field of view appropriate to the patient’s size. The CT examination number and the milliamperage and reconstruction field of view used at the examination were recorded. The images obtained at the conventional tube current chest CT examinations were then archived onto optical disks.
Simulated Scans
To create the simulated reduced tube current scans, the raw scan data from each of the 150 conventional tube current diagnostic examinations were extracted from the scanner database, written onto 8-mm magnetic tape, and transported to a workstation (Silicon Graphics O2; Silicon Graphics, Mountain View, Calif). The signals in every projection of the raw scan data were modified by the addition of computer-simulated Poisson-distributed random noise by using a custom program written for the numerical analysis language PV-WAVE (Visual Numerics, Boulder, Colo). Only the level of random noise was modified. The mean signal level was not affected by this transformation. The amount of noise added was adjusted to simulate reduction in the tube current from the conventional tube current value to either 100 or 40 mA. The scanner manufacturer (GE Medical Systems) provided all the computer programs that were used to manipulate the raw scan data. This noise-modifying technique was identical to that previously validated by Mayo et al (12).
The modified raw scan data (consisting of a simulated 100-mA acquisition and a simulated 40-mA acquisition for each conventional tube current CT examination included in this study) were transported back to the CT scanner and added to the raw scan database. The modified raw scan data sets were reconstructed by using standard and high-spatial-frequency algorithms to create mediastinal images and lung images, respectively, with the same field of view as that used at the conventional tube current CT examination.
The conventional and simulated reduced tube current scans were photographed by using mediastinal (window width, 450 HU; window level, 35 HU) and lung (window width, 1500 HU; window level, –700 HU) settings on a laser multiformat camera (Agfa Imaging, Greenville, SC). All technical annotations and identifying marks were removed. The resultant 450 complete chest CT image data sets (ie, 150 image data sets obtained with conventional tube current [200–320 mA], 150 obtained with the simulated reduced tube current of 100 mA, and 150 obtained with the simulated reduced tube current of 40 mA) were entered into a database.
Scan Interpretation
The 450 image sets were then assigned random numbers and separated into 10 groups of 45 image sets each for interpretation. Each group of 45 image sets included 15 conventional tube current image sets, 15 simulated 100-mA image sets, and 15 simulated 40-mA image sets. The 15 image sets at each milliamperage level were selected with the criterion that the same patient was represented only once in any group of 45 image sets. Four radiologists (in alphabetical order, T.J., P.K., K.I.K., S.L.S.M.) with 1–2 years of postresidency subspecialty training in chest radiology independently interpreted the mediastinal and lung window images of each examination. The observers were blinded to the tube current composition of the 10 groups. Because images were interpreted on film, there was no manipulation of window width or level settings. Depending on individual clinical workloads, observers took 2–4 months to score the entire collection of 450 image sets.
For each image set, the observers viewed all of the mediastinal and lung window images together. Images were viewed in a darkened room by using standard view boxes, replicating the interpretation technique used clinically. A standardized scoring sheet was provided.
Seven structures—the aorta, heart, esophagus, pleura, chest wall, mediastinum, and hilar nodes—were classified as normal or abnormal. Three of the structures were subcategorized as follows: The pleura was categorized as showing mass, effusion, thickening, or pneumothorax; the chest wall, as showing a bone lesion or other lesion; and the mediastinum, as showing lymphadenopathy or other disease. Seven lung findings were assessed. The observers noted the presence or absence of emphysema, small lung nodules (<1 cm), large lung nodules (1–3 cm), lung masses (>3 cm), atelectasis, consolidation, and linear scarring. Atelectasis was differentiated from linear scarring by the presence of volume loss. Small nodules, large nodules, and masses were subcategorized as being calcified or noncalcified.
Observers scored their subjective impression of mediastinal and lung window image quality by using a five-point scale in which a score of 1 meant nondiagnostic; a score of 2, inferior; a score of 3, adequate; a score of 4, good; and a score of 5, excellent.
After each observer had interpreted the results of all 450 examinations in the data set, the 150 conventional tube current image sets were separated from the 100- and 40-mA simulated reduced tube current image sets. After a minimum interval of 3 weeks, the 150 conventional tube current image sets were reinterpreted in random order by the same four radiologists with the same scoring system for structures and lung findings. This repeat interpretation of the conventional tube current image sets was performed in four to six sessions over 2–6 weeks, depending on the workload of the individual observers. Assessment of subjective image quality was not repeated because at the second reading the observers were aware of the tube current used to obtain the images.
Data Analysis
In all cases, the analysis involved comparing each observer’s performance with his or her performance at another time. The data for all 14 structures and lung findings for each observer were entered into a spreadsheet computer program (Excel; Microsoft, Redmond, Wash). The data obtained at the initial interpretation of the conventional tube current image sets were used as the reference standard to which the data obtained at the other three interpretations (the interpretation of the 100-mA image sets, the interpretation of the 40-mA image sets, and the second interpretation of the conventional tube current image sets) were compared. The overall objective of the analysis was to determine whether agreement between the interpretations of the reduced tube current image sets and the reference-standard interpretations differed from agreement between the repeat (second) interpretations of the conventional tube current image sets and the reference-standard interpretations.
First, overall agreement between the reference-standard and second interpretations of the conventional tube current image sets was determined for each observer. Overall agreement for each case required agreement for all 14 structures and lung findings, including subcategories. The presence or absence of overall agreement between the reference-standard interpretations and the interpretations of the 100- and 40-mA reduced tube current image sets was then determined. Data from the observers were pooled. An agreement analysis was then performed for each of the 14 structures and lung findings, including subcategories, to determine which categories were most responsible for adversely affecting agreement due to the use of reduced tube current.
The performance of reduced tube current scan interpretations relative to the repeat conventional tube current scan interpretations was formally examined by constructing a series of 2 x 2 tables. Within each table (Table 1), agreement between the reference-standard and repeat interpretations of the conventional tube current scans is reflected by the row margins, and agreement between the reference-standard and the reduced tube current (either 100 or 40 mA) scan interpretations is reflected by the column margins. The row margin totals identify the number of times the repeat interpretation of the conventional tube current scans agreed or disagreed with the reference-standard interpretation. The column margin totals identify the number of times the interpretation of the reduced tube current scans agreed or disagreed with the reference-standard interpretation.
|
Statistical Analysis
The McNemar test for paired data (13), yielding a 
2 statistic with one degree of freedom, was used as a test of differences between the off-diagonal elements in the 2 x 2 tables. The association between the subjective rating of the image quality (graded on a scale of 1–5) of mediastinal and lung window images according to tube current (conventional vs 100 mA, and 100 mA vs 40 mA) was tested for each observer by using the Wilcoxon matched-pairs signed rank test. The correlation between the conventional milliamperage value selected by the technologist (200–320 mA) and patient weight was tested by using the Pearson correlation coefficient. Statistical analyses were performed by using the SPSS/PC+ version 4.0 (SPSS, Chicago, Ill) and S-Plus version 6.1 for Windows (Insightful, Seattle, Wash) software. P < .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Averaged over the four observers, the rate at which at least one abnormal structure or lung finding in the conventional tube current image sets was reported was 95% (483 of 510 sets). On average, pleural abnormalities were the most frequent abnormality found by using mediastinal settings, while linear scarring was the most frequent abnormality detected by using lung settings. Esophageal abnormalities were the least frequent finding; they were seen in only 8% of cases (41 of 510 sets). For the three observers who read images from the entire collection of 150 examinations, there was the potential to report 1050 structures as abnormal (ie, 7 x 150). There was the same number of potential lung findings (ie, 1050). Observer 1, who read only 60 cases, had the potential to score 420 (ie, 7 x 60) abnormal structures and lung findings. A large difference in the rate of reporting abnormal structures and lung findings was seen between the four observers (Table 2), reflecting differences in individual scoring thresholds.
|
|
|
Table 5 shows the percentage of overall agreement between the initial and repeat readings of the reference-standard scans for each of the four observers. These values are derived from the row totals for the individual readers in the overall agreement tables (E1 and E2, radiology.rsnajnls.org/cgi/content/full/2323030899/DC1). These data demonstrate the variations in the level of overall agreement for two interpretations of the same images: those obtained at the conventional tube current CT examination. These values range from 80% (120 of 150 cases) for reader 4 to 17% (26 of 150 cases) for reader 3. Reader 3 showed substantially lower agreement at repeat reading of the conventional tube current scans than did the other three observers. As such, this reader represents an outlier compared with the other observers, and data from this reader were not included in the analysis described below.
|
These tables show greater agreement between the initial and the repeat reference-standard readings than between the readings of the reduced tube current scans and the initial reference-standard readings for all categories—that is, all structures and lung findings contributed to disagreements between the interpretations of the conventional and the reduced tube current scans. A significant (P < .05) number of disagreements between the readings of the conventional tube current scans and the readings of the reduced tube current scans occurred at both 100 and 40 mA for findings relating to the esophagus, pleura, chest wall (Fig 1), hilar nodes, emphysema, nodules 10–30 mm in size, and atelectasis. A significant difference (P < .05) was also seen at 100 mA for nodules smaller than 10 mm (Fig 2) and at 40 mA for the mediastinum and scar categories. The number of disagreements was not significant for findings related to the aorta, heart, presence of a mass larger than 30 mm, or consolidation.
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Using this technique to investigate the effect of reduced tube current on reader evaluation of 14 structures and lung findings, we found that as the tube current is reduced, overall agreement between interpretations of the reduced tube current scans and interpretation of the original conventional tube current scans decreases. A significant number of disagreements (P < .05) were seen when all observations were pooled, and—for three of four observers at 100 mA and two of four observers at 40 mA—when the results of individual observers were analyzed separately.
The finding that reduced tube current CT examinations result in degradation of the interpretation of diagnostically significant findings differs from the results of previous studies by Lee et al (11), Rusinek et al (14), and Mayo et al (10). However, these previous studies differ substantially from the current investigation with regard to the number of CT sections analyzed, the CT technique used, the outcome measured, and the sample size.
Lee et al (11) found no significant difference (P > .05) when comparing the diagnostic accuracy of 340-mA tube current thin-section CT scanning with that of 80-mA tube current thin-section CT scanning in 50 cases of chronic interstitial lung disease. There were substantial differences between that study and the current study. First, the 1-mm-collimation thin-section CT images used by Lee et al have a much higher level of image noise, even at 340 mA, than the helical 7-mm (with a pitch of 1) images used in the current study. The higher noise level is a result of the 1-mm collimation, which reduces the x-ray photon flux by a factor of seven.
Second, the study of Lee et al involved assessment of the effect of reduced tube current thin-section CT on the diagnostic accuracy of five categories of chronic interstitial lung disease in 50 patients and 10 control subjects. The overall pattern and distribution of disease contribute substantially to the differential diagnosis of chronic interstitial lung disease at thin-section CT (this disease can frequently be assessed on a limited number of sections), in contrast to the present analysis of multiple structures and lung findings (14 in this study) required to interpret conventional helical CT examination results in patients with a wide variety of chest problems.
The conclusion drawn in the present study also differs from that drawn in the study of Rusinek et al (14), who quantified the effectiveness of reduced tube current CT for the detection of simulated pulmonary nodules. Because the lung nodules were simulated, a true standard of reference, the number of simulated lung nodules, could be applied. The presence of an external standard of reference allowed Rusinek et al to calculate the sensitivity and specificity for reduced tube current CT examinations. Their data indicated that there is no significant decrease in sensitivity between 200 mAs (63%) and 20 mAs (60%) in the overall detection of lung nodules. Therefore, they concluded that screening CT for the detection of lung nodules could be performed with reduced tube current.
In the current study, lung nodules represented only two of the 14 structures and lung findings assessed. However, the finding that the detection of lung nodules smaller than 1 cm and of those 1–3 cm in size was a source of disagreement in the present study suggests that once a nodule is found, a diagnostic chest CT examination should be performed by using a higher tube current to optimally characterize the lung nodule and identify any other nodules or chest abnormalities. This is in agreement with current clinical practice (15,16).
The conclusion of the current study also differs from the conclusion in the study of Mayo et al (10), who found no significant difference (P > .05) in the detection of three mediastinal and two lung abnormalities in 30 subjects who underwent reduced tube current CT examinations. In that study, scans obtained with a standard tube current of 400 mAs were compared with scans obtained with reduced tube current (200, 140, 80, 20 mAs) at two levels in the chest (the tracheal carina and left atrium). The small number of diagnostic categories and subjects in that study limited its statistical power; this may have influenced the conclusion. The current study is much larger in terms of both the number of categories and the number of subjects studied, which enhanced its ability to reveal differences between the levels of tube current. Additionally, the simulation technique employed in this study allowed the generation of complete reduced tube current examinations, increasing the number of abnormalities detected per subject. These factors all increased the statistical power of this study.
Our observation that a correlation exists between decreasing tube current and decreasing subjective image quality is in agreement with results of a previous study (10) that involved use of a repeated scanning protocol at two levels. This correlation supports the hypothesis that the simulated reduced tube current images "behave" like genuine reduced tube current images. The weak but significant correlation between subject weight and the tube current selected by the technologist for the conventional tube current CT examination suggests that this technique of tube current adjustment is inconsistently applied in clinical practice.
There were several potential limitations of this study. The study was limited by the absence of an external standard of reference. In the absence of external validation, we used the first interpretation of the conventional tube current scans as the reference standard. We then compared the reproducibility of repeated interpretation of the conventional tube current scans with the interpretation of the reduced tube current scans. We argue that if the reduced tube current scans provide the same information as the conventional tube current scans, and if observers are consistent in identifying and recording structures and lung findings, there should be no significant difference in terms of the level of agreement between the initial and repeat reference-standard readings and the level of agreement between the initial reference-standard reading and the reduced tube current scan reading.
Alternatively, if the performance of the reduced tube current scan interpretation differs from that of the repeat interpretation of the conventional tube current scans, as measured with the McNemar test, that difference is attributed to the effect of reduced tube current. The McNemar test reveals statistically significant disagreement caused by either unusually high disagreement between the initial and repeat reference-standard readings or between the initial reference-standard and the reduced dose scan readings. In only one instance (reader 3, pooled structures and lung findings at 40 mA) was the significant (P < .05) excess disagreement caused by greater disagreement between the initial and repeat reference-standard readings than between the reference-standard and reduced tube current readings. This may reflect the fact that reader 3 is able to accurately interpret reduced tube current scans despite the presence of increased levels of noise. Alternatively, and more likely in view of the excess of disagreements between the first and second interpretations of the conventional tube current scans for this reader, it may be due to inconsistent classification of CT findings.
For the purpose of the study, intraobserver agreement required consistent classification of a total of 14 mediastinal and lung abnormalities (including subgroup classifications) rather than agreement as to the final diagnosis. Because reader 3 showed only 17% intraobserver agreement at repeat reading of the conventional tube current scans, this reader’s data were not included in the analysis of structures and lung findings that contributed to disagreements at reduced tube current levels.
The order in which the conventional tube current, 100-mA, and 40-mA scans were read was not recorded, making it impossible to assess the effect of reading order on the data. Therefore, we cannot determine if the memory of a particular abnormality at one interpretation of a scan influenced its detection when the scan of the same patient at a different tube current was interpreted. Additionally, although two scans obtained in the same patient at the same tube current were never present in any group of 45 image sets, we did not ensure that the scans of any one patient were separated by a specific time interval. Therefore, we relied on the sheer volume of interpretations to minimize the effect of observer memory on the results.
We note that the second reading of the conventional tube current image sets always followed these previous three interpretations. This may have created a learning bias in which the observers became familiar with abnormalities as the scans were repeatedly interpreted. The influence of this effect should be to increase the level of disagreement in the repeat reading of the conventional tube current image sets. The fact that more disagreements were seen between interpretations of the reference-standard scans and interpretations of the reduced tube current scans than between the initial and repeat readings of the conventional tube current scans supports the hypothesis that the reduced tube current increased reader disagreement with the reference-standard scans.
The large number of scans included in this study accounts for the fact that reader 1 only completed interpretations at each tube current and repeated interpretation of the conventional tube current scans for 60 of 150 subjects. However, the experimental design, which involved comparison of each observer’s data with his or her own data, makes it possible to add the results of reader 1 to the larger data sets of readers 2, 3, and 4.
Images in this study were acquired by using single-section helical CT scanners and interpreted on film at a fixed window and level setting. Further studies would be necessary to assess the effect of reduced tube current on the interpretation of images acquired with multi–detector row CT scanners and read at workstations by using operator-controlled window width and level settings.
In summary, the results of this study show that as the tube current is reduced at helical CT examinations, the level of agreement with the reference-standard conventional tube current scans decreases relative to that of replicate readings of the reference-standard scans. Therefore, information regarding structures and lung findings is degraded. This may contribute to a decrease in diagnostic accuracy, but this was not assessed in this study. These data suggest that strategies for radiation dose reduction at helical chest CT should be directed toward techniques that reduce radiation exposure without substantially increasing image noise. One example of such a technique is reduction of the tube current depending on the thickness of the body as viewed in a given projection (17,18). The results of this study suggest that radiation dose reduction based on simple tube current reduction could potentially affect the diagnostic accuracy of interpretation of helical CT examination results.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, J.R.M., S.V.; study concepts and design, J.R.M., S.V.; literature research, K.I.K., J.R.M.; clinical studies, K.I.K., T.J., P.K.; experimental studies, K.I.K., H.O.C.; data acquisition, K.I.K., T.J., P.K., H.O.C., S.L.S.M.; data analysis/interpretation, H.O.C., S.V., J.R.M.; statistical analysis, S.V.; manuscript preparation, K.I.K., S.V., J.R.M., S.L.S.M.; manuscript definition of intellectual content and final version approval, J.R.M., S.V.; manuscript editing, H.O.C., S.L.S.M.; manuscript revision/review, H.O.C., S.V., S.L.S.M.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. Kubo, P.-J. P. Lin, W. Stiller, M. Takahashi, H.-U. Kauczor, Y. Ohno, and H. Hatabu Radiation Dose Reduction in Chest CT: A Review Am. J. Roentgenol., February 1, 2008; 190(2): 335 - 343. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. MacKenzie, J. Nazario-Larrieu, T. Cai, M. S. Ledbetter, M. A. Duran-Mendicuti, P. F. Judy, and F. J. Rybicki Reduced-Dose CT: Effect on Reader Evaluation in Detection of Pulmonary Embolism Am. J. Roentgenol., December 1, 2007; 189(6): 1371 - 1379. [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]
|
|
|
|
|
|
D. K. Yousefzadeh, M. B. Ward, and C. Reft Internal Barium Shielding to Minimize Fetal Irradiation in Spiral Chest CT: A Phantom Simulation Experiment. Radiology, June 1, 2006; 239(3): 751 - 758. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. S. Parker, F. K. Hui, M. A. Camacho, J. K. Chung, D. W. Broga, and N. N. Sethi Female Breast Radiation Exposure During CT Pulmonary Angiography Am. J. Roentgenol., November 1, 2005; 185(5): 1228 - 1233. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
