| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 All authors: Mallinckrodt Institute of Radiology, Washington University School of Medicine, 510 S Kingshighway Blvd., St. Louis, MO 63110.
Received June 17, 2003;
accepted after revision September 16, 2003.
Address correspondence to M. J. Siegel
(siegelm{at}mir.wustl.edu).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Fourteen patients, 17 days to 20 years old, with thoracic aortic anomalies underwent MDCT using axial, multiplanar, and 3D volume-rendering imaging. All images were reviewed by three radiologists for position of the aortic arch, coarctation, vascular compression of the airway, collateral vessel formation, and aortopulmonary shunts (patent ductus arteriosus). Final diagnosis was determined by echocardiography, conventional angiography, bronchoscopy, or surgery. Diagnostic accuracy, sensitivity, and interobserver agreement were evaluated.
RESULTS. Average accuracies (average of the three observers for a correct diagnosis) were greater than or equal to 96% for diagnoses of aortic position and airway narrowing on all image types. For the diagnosis of coarctation, average sensitivities (average of the three observers for a true diagnosis) were 73% for axial, 100% for multiplanar, and 100% for 3D volume-rendered images. For the diagnosis of patent ductus arteriosus, average sensitivities were 78% for axial, 94% for multiplanar, and 89% for 3D volume-rendered images. No patients in this study had collateral vessel formation. For the diagnosis of absence of collateral vessel formation, average sensitivities were 100% for axial, 100% for multiplanar, and 100% for 3D volume-rendered images. There were no significant statistical differences in diagnostic performances, agreement with truth, or confidence scores among observers or imaging formats (p > 0.05).
CONCLUSION. Axial, multiplanar, and 3D volume-rendered images serve equally well as methods for assessing the side of the aorta to diagnose anomalies. For evaluation of coarctation and patent ductus arteriosus, multiplanar and 3D volume-rendered images perform slightly better than axial images.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Sedation
Seven pediatric patients, who ranged in age from 1 month to 2 years, were
sedated. Orally administered chloral hydrate (50–100 mg/kg; maximum
dose, 2,000 mg) was used in patients younger than 18 months and IV
administered pentobarbital sodium (6 mg/kg; maximum dose, 200 mg) in those
older than 18 months. The remaining seven patients cooperated without
sedation.
Imaging Technique
Contrast enhancement factors.—MDCT was performed with either
a Somatom Plus 4 scanner or Sensation 16 scanner (Siemens Medical Systems,
Iselin, NJ). All examinations were contrast-enhanced. Patients received 2
mL/kg (not to exceed 125 mL) of nonionic contrast medium (320 mg I/mL). In
patients with antecubital catheters, contrast medium was injected with a power
injector at a rate of 1.5–2.0 mL/sec for a 22-gauge catheter and 3
mL/sec for a 20-gauge catheter. Manual injection was used when IV access was
via a catheter placed in the dorsum of the hand or wrist. Scanning was
initiated 12–15 sec after the start of the IV contrast medium injection
in patients weighing 15 kg or less. For patients weighing more than 15 kg,
scanning initiation was triggered by an automated tracking system. The target
vessel for automated tracking was the ascending aorta. The threshold level was
100 H. A single bolus of contrast material was administered for all
examinations.
Technical factors.—All CT examinations were performed with low radiation dose techniques. Studies were performed at 25 mAs in patients weighing less than 15 kg, at 30 mAs in patients weighing between 15 and 24 kg, at 45 mAs in patients weighing between 25 and 34 kg, at 75 mAs in patients weighing between 35 and 44 kg, at 100 mAs in patients weighing between 45 and 54 kg, and at 120–140 mAs in patients weighing more than 54 kg. An 80-kV dosage was used for patients weighing less than 50 kg and 120 kV for patients weighing more than 50 kg.
CT examinations on 4-MDCT scanners were performed with 2.5-mm collimation and a table speed of 15–20 mm per rotation. Images were reconstructed at 5-mm thickness for routine viewing and at 3-mm thickness for multiplanar and 3D reconstruction. On the 16-MDCT scanner, CT examinations were performed with 1.5-mm collimation and a table feed of 24–36 mm per rotation. Images were reconstructed for viewing at 5 mm and at 2 mm for multiplanar and 3D reconstructions, respectively. Section intervals were 1 mm for all 3D reconstructions.
Scans extended from just below the level of the thoracic inlet, so that the proximal aspects of the common carotid and subclavian arteries were included, to the level of the renal arteries. In sedated patients and patients younger than 5 years, imaging was performed during quiet respiration. Patients 5 years or older were able to suspend respiration for the duration of the scanning. The time for the entire CT examination was less than 15 sec on the 4-MDCT scanner and less than 6 sec on the 16-MDCT scanner.
All images were processed with standard soft-tissue (e.g., width, 400–450 H; level, 40–50 H) and lung (e.g., width –1,600 to –1,800 H; level, –450 to –550 H) window settings.
Image processing.—Reviewers had access to unlimited multiplanar images, reformatted at 1-mm intervals. These were generated at the workstation during the review session. Sagittal, coronal, and oblique multiplanar images were reviewed for all cases.
We used volume-rendering algorithms to highlight the vascular and airway structures. The reconstructed images were generated at a separate workstation (Vitrea 2 workstation, Vital Images, Plymouth, MN) before the interpretation session by a radiologist and technologist who were experienced in 3D reconstructions and unaware of the final diagnosis. They were able to complete the reconstructions in 30 min. The images were selected so that the aorta and its branches, pulmonary arteries, pulmonary veins, superior vena cava and other thoracic veins, and the trachea and its branches were always seen and displayed in a similar fashion for all cases. The 3D reconstructed images were presented to the observers in two formats—one with more opacity to show the vessels and the other with more transparency to show the airway. Contiguous osseous structures were edited from the reconstruction volume. Patient identifying information was removed from the images. This approach ensured that the reconstructions were uniform for all 14 cases. The observers reviewed the 3D reconstructed images in real time and thus could rotate the images to observe every vessel from any perspective.
Interpretation of Studies
Axial, multiplanar, and 3D volume-rendered images were evaluated
independently and retrospectively by three radiologists. One reviewer was an
attending chest radiologist. The second reviewer was an attending pediatric
radiologist. Both have more than 15 years' experience in interpreting CT
examinations. The third reviewer was a chest radiologist who just finished 1
year of chest radiology fellowship. The reviewers were unaware of the results
of other imaging techniques, surgery, or bronchoscopy. The axial CT images
were interpreted first, followed by the multiplanar images at the same review
session. The 3D volume-rendered images were interpreted independently at a
later session. The order of review was randomized, and the interpretation
session for the 3D volume-rendered images was separated in time by at least 2
weeks to avoid recall of patient diagnosis.
All image sets were evaluated for aortic position, coarctation, vascular compression of the airway, collateral vessel formation, and aortopulmonary shunts (i.e., patent ductus arteriosus). Aortic coarctation was defined as greater than 25% decrease in vessel diameter, either focal or diffuse. Airway narrowing was defined as greater than 25% decrease in tracheal diameter, either focal or diffuse. Criteria for diagnosis of collateral vessel formation varied with age. The mammary and intercostal arteries are virtually never seen in neonates and infants younger than 1 year. Thus, we considered visualization of these structures in this age group to be abnormal. In older infants and children, the vessels were considered abnormal if they were tortuous or their diameter exceeded that of the coronary arteries. Published standards were used in adolescents and adults [7]. In this age group, a diameter exceeding 3.5 mm was considered abnormal. Patent ductus arteriosus was defined as a tubular connection between the proximal descending aorta and main, or left, pulmonary artery. The data were recorded on an imaging evaluation form.
Reviewers scored their confidences in diagnoses with a 5-point confidence scale from 1 (least confident) to 5 (most confident).
Final Diagnoses
The medical records were reviewed to determine the truth of diagnosis. The
reference standard for the diagnosis of vascular lesions was echocardiography
in six patients, angiography in one, and both echocardiography and angiography
in six. Surgical confirmation also was available in 10 of these 13 patients.
One patient with a double aortic arch was not evaluated with further imaging
study or surgery because the CT findings of double aortic arch explained
abnormal findings on chest radiography. The standard of reference for presence
or absence of airway narrowing was bronchoscopy in four patients, surgery
performed for correction of aortic abnormalities in six, and clinical history
in four. Angiography, echocardiography, bronchoscopy, and surgery were
performed by experienced pediatric subspecialists.
Statistical Analysis
Four of the five categories had binomial data (i.e., yes or no). For these
four categories, differences in diagnostic accuracy among reviewers and sets
of images were tested with McNemar tests, and sensitivities were calculated.
In the one category (side of the aortic arch) that had multiple nominal
categories (right, left, or cervical), differences in diagnostic accuracy
among reviewers and sets of images were assessed with Cohen's kappa statistic.
Sensitivities could not be calculated because of the multiple nominal
categories. Marginal homogeneity tests for ordered data were used to test for
differences in observer confidence scores. Weighted kappa statistic was used
as a measurement of agreement between confidence scores. Alpha was set at
0.05. Statistical analysis was performed with StatXact 5 statistical software
(Cytel Software, Cambridge, MA).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Four patients had abnormalities in arch position. Patient ages ranged from 1 month to 20 years (mean, 7.7 years). A right aortic arch with an aberrant left subclavian artery was present on CT in one patient and confirmed surgically (Fig. 1A, 1B, 1C, 1D), and a ligamentum arteriosum was found at surgery. A right aortic arch with mirror image branching was present in one patient who had undergone prior repair of tetralogy of Fallot. A diagnosis of double aortic arch was made on CT in another patient (Fig. 2A, 2B, 2C, 2D). A cervical arch was documented on CT and on conventional angiography in one other patient.
|
|
|
|
|
|
|
|
Tracheal narrowing (> 25% decrease in luminal diameter) was identified in the two patients with vascular rings (right arch with aberrant left subclavian artery and double arch). Results were confirmed at bronchoscopy. Neither patient was symptomatic.
Five patients had shunt lesions. Patient ages ranged from 1 month to 17 years (mean, 4.4 years). Patent ductus arteriosus was shown on conventional angiography in three patients and on echocardiography in two patients. The diameters of the patent ductus arteriosus ranged from 5 to 7.5 mm.
No patients in this study had collateral vessel formation. This likely reflects the young age of the patients with coarctations.
Statistical Analysis
For all determinations, observers averaged three misses with axial imaging,
1.7 misses with 3D volume-rendered imaging, and 0.3 misses with multiplanar
imaging (Table 1). The average
accuracies for all lesions were 96% (201/210) for axial imaging, greater than
99% (209/210) for multiplanar imaging, and 98% (205/210) for 3D
volume-rendered imaging. With axial imaging, 188 (90%) of 210 confidence
ratings were 5. With multiplanar imaging, 210 (100%) of 210 confidence level
ratings were 5, and with 3D volume-rendered imaging, 207 (99%) of 210 were 5
(Table 2). Specific details of
the statistical analyses follow.
|
|
Side of arch.—Three arches were documented as right, nine as left, one as cervical, and one as double on studies used as the truth standard. With multiplanar and 3D volume-rendered images, there was perfect agreement among observers and with the truth standard; average accuracies were 100%. With axial images, accuracies were 96%. One observer misdiagnosed the cervical arch as a normally positioned left arch on axial images. For this observer, the kappa value for agreement with the truth standard and other observers was 0.96 (± 0.05, SE). Confidence scores were 5 except for one 4 for a cervical arch on axial images by one observer. There was, therefore, no significant difference in confidence scores among observers (p > 0.99). Because of zero frequency cells, kappa for agreement for confidence scores could not be calculated.
Coarctation.—Five vessels were identified as narrowed and nine as normal on studies used to document the truth standard. With multiplanar and 3D volume-rendered images, there was near-perfect agreement among observers and with the truth standard (p > 0.99). Axial images had the worst performance for vessel narrowing. Average accuracies were 90% for axial, 100% for multiplanar, and 100% for 3D volume-rendered images. Sensitivities were 73% for axial, 100% for multiplanar, and 100% for 3D volume-rendered images. On axial imaging, two observers misdiagnosed the same two cases of coarctation as normal. Both the missed coarctations were approximately 5 mm in length (Fig. 3A, 3B, 3C, 3D). Confidence scores were 5 in 86% of cases on axial images, in 100% on multiplanar images, and in 93% on 3D volume-rendered images. Both observers had a confidence level of 4 for one of the cases missed on axial imaging, and one observer had a confidence level of 4 and one had a confidence level of 3 for the other case. Confidence levels were not, however, significantly different among observers (p > 0.99), and kappa was high (0.84 ± 0.14).
|
|
|
|
Airway compression.—Tracheal narrowing was documented on bronchoscopy in two cases. Narrowing was considered absent in 12 cases on the basis of clinical findings or surgery. There was perfect agreement among all observers with all imaging types and perfect agreement with truth (accuracies and sensitivities, 100%). In all cases, confidence scores were 5 on multiplanar and 3D volume-rendered images (Fig. 4A, 4B, 4C). On axial images, confidence scores were 5 in 90% of cases and 4 in 10% of cases. Two observers used a score of 4 in one case and one observer used a score of 4 in two cases. In the worst instance, in which one observer used two confidence scores of 4 compared with the confidence scores of 5 for multiplanar and 3D volume-rendered images, there was no significant difference (p = 0.25). Because of zero frequency cells, kappa for agreement for confidence scores could not be calculated.
|
|
|
Collaterals.—No patients in this study had collateral vessel formation documented on any of the reference studies used as truth. No observer diagnosed collateral vessel formation with any of the imaging types, and all confidence scores were 5.
Shunt lesions.—Five patients had patent ductus arteriosus
documented on angiography or echocardiography. For these diagnoses, average
accuracies were 90% for axial, 98% for multiplanar, and 95% for 3D
volume-rendered images (p > 0.05). Average sensitivities were 78%
for axial, 94% for multiplanar, and 89% for 3D volume-rendered images. On
axial images, all three observers missed a small patent ductus arteriosus in
the same patient (Fig. 5A,
5B,
5C), and one observer missed
an additional patent ductus arteriosus. On multiplanar images, one observer
missed a patent ductus arteriosus, and on 3D volume-rendered images, two
observers each missed a patent ductus arteriosus in the same patient. All
missed patent ductus arteriosi were 5 mm in diameter. All confidence scores
for multiplanar and 3D volume-rendered images were 5. Confidence scores for
axial images were 5 in 74% of cases. One observer used five scores of 4,
another used four scores of 4, and the last observer used two scores of 4. All
missed cases had confidence level scores of 5. For the worst case, in which
the observer missed two patent ductus arteriosi with axial imaging, there was
no significant difference from the truth standard or from the performance of
other observers with axial imaging (p = 0.50; 
= 0.70 ±
0.19). In the worst case in which five scores of 4 were used by one observer
for axial imaging, these scores did not differ significantly from the
confidence scores for which responses were 5 (p = 0.06). Because of
zero frequency cells, kappa for agreement for confidence scores could not be
calculated.
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Another difference between our study and the prior series was the use of the volume-rendering technique for generating 3D images. Prior series have almost exclusively used a 3D shaded-surface display, which displays only a subset of the volumetric data by including pixels in a certain range of attenuation values. Thus, large amounts of data are unused in the final reconstruction. Radiologists using the shaded-surface technique, by selecting threshold levels, also can potentially alter apparent diameters of vascular or airway lumina. The 3D volume-rendering technique uses all the information in the volumetric data set for the final reconstruction, and the 3D volume-rendered images are interactively displayed and can be edited in real-time [2, 3, 8, 9]. Moreover, the opacity and shading characteristics can be altered to display different types of tissues (i.e., vascular structures, airway, and soft tissues). Studies in adults have reported that volume-rendering techniques allow more accurate assessment of vascular stenoses [10].
The results of our study indicate that axial, multiplanar, and 3D volume-rendered images perform equally well in diagnosing the side of the aorta in pediatric patients and young adults. Sensitivity and confidence in diagnosis were high, exceeding 96%. A statistically significant interobserver agreement also was found. However, in the evaluation of coarctation and shunt lesions, multiplanar and 3D volume-rendered images performed slightly better than axial images. For the diagnosis of coarctation, average sensitivities were 73% for axial, 100% for multiplanar, and 100% for 3D volume-rendered images (p > 0.05). For the diagnosis of shunt lesions, average sensitivities were 78% for axial, 94% for multiplanar, and 89% for 3D volume-rendered images (p > 0.05).
The ability to visualize a vascular structure depends on the angle between these structures and the imaging plane. If a vessel is perpendicular to the longitudinal axis of the imaging plane, the cross-sectional diameter can be reliably displayed and stenosis can be more easily assessed. If the vessel is oblique or parallel to the imaging plane, stenoses can be missed. The high accuracy of all image types in the diagnosis of arch position is then understandable. The aorta courses perpendicular to the imaging plane and extends over several levels and, therefore, can easily be identified on axial imaging. The underdiagnosis of coarctation and patent ductus arteriosus on axial images is not that surprising, because these abnormalities are often small and they course obliquely rather than perpendicular to the imaging plane. For example, a short (5 mm) coarctation was misdiagnosed in two patients on axial images. On sagittal multiplanar and 3D volume-rendered images, these short coarctations were well visualized, leading to a correct diagnosis.
Because multiplanar images are easy to generate, we believe that they should be used to complement axial images in diagnosing vessel narrowing and shunt lesions. Three-dimensional volume-rendered images should also be generated. Manual editing of the 3D volume-rendered images in our experience required approximately 30 min at the workstation. By viewing the reformatted multiplanar or 3D volume-rendered images in oblique and sagittal planes, the radiologist can determine with more confidence whether short segment coarctation or a small shunt, which courses obliquely through the image set, is present. These lesions are the ones that are more likely to be missed in the axial plane.
Both patients with vascular rings in this study had airway narrowing, which was seen on all three imaging planes. Similar to the great vessels, the perpendicular and long course of the airway and the long-segment tracheal narrowing make imaging in the axial plane suitable. No patient in our study had bronchial narrowing. Oblique coronal and sagittal multiplanar and 3D volume-rendered images have been reported to be valuable in visualization of bronchial narrowing [2, 11, 12].
One limitation of our study was the small number of patients who had vascular or airway anomalies, which limited the statistical power of the evaluation. Even given this weakness, our results are important and point out potential errors in diagnosis, particularly in coarctation and patent ductus arteriosus. Another limitation was that some patients had only echocardiography for a standard of reference and did not undergo angiography. However, this comparison would have required more radiation exposure in pediatric patients than we could have ethically justified. Similarly, bronchoscopy, because of the sedation-related risks and additional costs, could not be justified.
Multiplanar images were interpreted after axial images at the same review session in our study. It is possible, therefore, that observer performance with multiplanar images may have benefited from recall of the axial images. We think, however, that this effect would have been small. For each of the 14 cases, there were five diagnostic tasks. Each observer would, therefore, have performed 70 diagnostic tasks with axial images before seeing the randomized multiplanar images. The observers would not, however, have known the truth standard; so recall even if it occurred would be of limited value.
Factors such as radiation exposure and sedation also need to be considered when CT is performed in an individual patient. Radiation exposure is directly proportional to tube current. Our study, in contrast to other studies evaluating helical CT in the assessment of mediastinal vessels and airways [4–6], used very low milliampere-seconds and decreased kilovoltage.
The results of this study suggest that low-dose MDCT is suitable for the diagnosis of congenital aortic anomalies in pediatric patients. Although CT has the disadvantage of ionizing radiation, the alternative studies also have inherent limitations. Echocardiography may be limited in its ability to define extracardiac vessels. MRI is time-consuming, may require prolonged patient sedation, and may be difficult to perform in seriously ill patients. Catheter angiography has a relatively high radiation exposure, may require long sedation times, and may have catheter access complications.
Because of faster imaging, MDCT also decreases the need for sedation in pediatric patients and thereby reduces imaging costs and resource usage [13]. Before the introduction of MDCT, almost 90% of children younger than 4 years were sedated at our hospital. In this study, 50% of patients younger than 4 years were sedated. Most sedations occurred early in our experience with 4-MDCT. With the use of the 16-MDCT scanner, the sedation rate has decreased to less than 10% in patients under 4 years old.
In conclusion, MDCT angiography can provide reliable diagnostic information about thoracic aortic anatomy in pediatric patients and young adults. Although axial images are diagnostic for evaluation of aortic position and vascular compression of the airway, they may be limited for evaluation of small coarctations and shunt lesions. In patients with these disorders, multiplanar and 3D volume-rendered images can enhance the diagnostic value of CT. High-quality diagnostic images can be obtained with relatively low radiation doses and decreased sedation rates.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. G. Eichhorn, C. Jourdan, S. L. Hill, S. V. Raman, J. P. Cheatham, and F. R. Long CT of Pediatric Vascular Stents Used to Treat Congenital Heart Disease Am. J. Roentgenol., May 1, 2008; 190(5): 1241 - 1246. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Ou, D. S Celermajer, G. Calcagni, F. Brunelle, D. Bonnet, and D. Sidi Three-dimensional CT scanning: a new diagnostic modality in congenital heart disease Heart, August 1, 2007; 93(8): 908 - 913. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B Oguz, M Haliloglu, and M Karcaaltincaba Paediatric multidetector CT angiography: spectrum of congenital thoracic vascular anomalies Br. J. Radiol., May 1, 2007; 80(953): 376 - 383. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z A Aziz, J C Davies, E W Alton, A U Wells, D M Geddes, and D M Hansell Computed tomography and cystic fibrosis: promises and problems Thorax, February 1, 2007; 62(2): 181 - 186. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E Y Lee, J E Dillon, M J Callahan, and S D Voss 3D multidetector CT angiographic evaluation of extralobar pulmonary sequestration with anomalous venous drainage into the left internal mammary vein in a paediatric patient. Br. J. Radiol., September 1, 2006; 79(945): e99 - e102. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Dodge-Khatami, S. Ott, S. D. Bernardo, and F. Berger Carotid-Subclavian Artery Index: New Echocardiographic Index to Detect Coarctation in Neonates and Infants Ann. Thorac. Surg., November 1, 2005; 80(5): 1652 - 1657. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. S. M. Chan, W. C. W. Chu, K. L. Cheung, A. A. Arifi, and W. W. M. Lam Angiography and Dynamic Airway Evaluation with MDCT in the Diagnosis of Double Aortic Arch Associated with Tracheomalacia Am. J. Roentgenol., November 1, 2005; 185(5): 1248 - 1251. [Full Text] [PDF]
|
|
|
|
 |
|
 M. I. Boechat, O. Ratib, P. L. Williams, A. S. Gomes, J. S. Child, and V. Allada Cardiac MR Imaging and MR Angiography for Assessment of Complex Tetralogy of Fallot and Pulmonary Atresia RadioGraphics, November 1, 2005; 25(6): 1535 - 1546. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Gralla, F. Spycher, C. Pignolet, C. Ozdoba, P. Vock, and H. Hoppe Evaluation of a 16-MDCT Scanner in an Emergency Department: Initial Clinical Experience and Workflow Analysis Am. J. Roentgenol., July 1, 2005; 185(1): 232 - 238. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. Backer, C. Mavroudis, C. K. Rigsby, and L. D. Holinger Trends in vascular ring surgery J. Thorac. Cardiovasc. Surg., June 1, 2005; 129(6): 1339 - 1347. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.-F. Paul, H. T. Abada, A. Sigal-Cinqualbre, E. Lee, and M. Siegel Should Low-Kilovoltage Chest CT Protocols Be the Rule for Pediatric Patients? Am. J. Roentgenol., October 1, 2004; 183(4): 1172 - 1172. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
