| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Developments |
1 From the Institute of Diagnostic Radiology, University Hospital Zurich, Switzerland (T.S., S.W., H.A., M.K., B.M., T.B.); and Institute of Radiology, Kantonsspital Chur, Spitäler Chur, Loestrasse 170, CH-7000 Chur, Switzerland (T.B.). Received November 2, 2004; revision requested January 1, 2005; revision received January 27; accepted February 25. Supported by the National Center of Competence and Research, Computer Aided and Image Guided Medical Interventions, of the Swiss National Science Foundation. Address correspondence to T.B. (e-mail: thomas_boehm{at}gmx.net).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and MethodsResultsDiscussionReferences |
|---|
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionReferences |
|---|
Four–detector row computed tomography (CT) has shown great potential for the noninvasive assessment of peripheral runoff vessels (3–6). However, one group of authors reported a decline in diagnostic concordance between CT angiography and DSA of the lower leg compared with the concordance between CT angiography and DSA of the aortoiliac and upper leg vascular tree (6). This might be due to the limited transverse resolution of four–detector row CT. Fast acquisition and isotropic volumetric scanning with 16–detector row CT scanners enable CT angiography to be performed with coverage of the abdominal aorta and the entire leg in a single continuous acquisition with submillimeter transverse resolution (7,8). Thus, the aim of our study was to prospectively assess the diagnostic accuracy of three reconstructions of 16–detector row CT angiographic data with different section widths and reconstruction increments (2.0 and 1.0 mm, 1.0 and 0.5 mm, and 0.75 and 0.4 mm, respectively) in patients with lower-leg PAD.
|
Materials and Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionReferences |
|---|
Underlying diseases did not play a role in referral or study enrollment; therefore, patients in whom diabetes mellitus was the underlying disease (n = 5) (ie, patients with a higher probability of developing calcified stenoses) were included with the same probability as patients with PAD but without diabetes. Four of these patients had non–insulin-dependent diabetes, and one patient had insulin-dependent diabetes. CT angiography was performed within 7 days after DSA. Exclusion criteria were a serum creatinine level of greater than 120 µmol/L and/or a previous adverse reaction to iodinated contrast medium. The study was approved by the review board of University Hospital Zurich, and written informed consent was obtained from all patients.
DSA Technique
DSA (the reference standard examination) was performed with a retrograde technique by a cardiovascular radiologist (T.B., with 10 years of experience in cardiovascular radiology) with a 1024 x 1024 image matrix angiography unit (Integris V5000; Philips Medical Systems, Best, the Netherlands). A 5-F pigtail catheter (AngiOptic; Angiodynamics, Queensbury, NY) was placed transfemorally in the suprarenal aorta for angiographic assessment of the aorta and renal arteries, and 30 mL of a nonionic iodinated contrast medium (iodixanol, Visipaque 320; Amersham Health, Buckinghamshire, England) was injected at a flow rate of 15 mL/sec. Then, the catheter tip was pulled down and positioned above the aortic bifurcation for imaging the arteries of the pelvis, thighs, knees, and lower limbs.
Angiography of the pelvic arteries was performed in a 30° oblique projection on both sides by using 14 mL of the contrast medium at a flow rate of 7 mL/sec. For angiographic assessment of the femoropopliteal and lower limb arteries, 14–30 mL of the contrast medium was injected at the same flow rate, and images were obtained in the anteroposterior projection. In cases of insufficient contrast enhancement owing to the presence of severe stenoses, side-selective catheter placement and contrast injection were performed. Intraarterial vasodilating drugs were not administered because diagnostically sufficient vessel opacification was achieved in all patients without such drugs. If asymmetric stenosis was suspected, additional oblique projections were obtained. Images from DSA were printed out as hard copies. For the current study, only the DSA images of the lower limb were used.
CT Angiography Technique
CT angiography was performed by using a 16–detector row CT scanner (Sensation 16; Siemens, Forchheim, Germany). The scanning range was planned with a scout view and included the entire vascular tree from the abdominal aorta down to the ankles. First, a single nonenhanced low-dose scan (20 mAs) of the abdominal aorta at the level of the origin of the renal arteries was obtained. On this image, a 10–15 mm2 region of interest (ROI) was placed in the lumen of the aorta for all 17 CT angiography studies by the same radiologist (T.S., with 5 years of experience). This ROI was used to help optimize the intraluminal contrast enhancement. The delay time from contrast material injection to scanning was determined individually and automatically for each patient by using a bolus-tracking technique (9).
A total of 150 mL of contrast material (iodixanol, Visipaque 320; Amersham Health) was administered with an automated injector (Ulrich Medical, Ulm-Jungingen, Germany) at a flow rate of 3 mL/sec through a 20–22-gauge needle that was placed in a superficial vein located in the antecubital fossa. At the same flow rate, 30 mL of saline was intravenously injected after the contrast material administration. After the preset contrast enhancement level of 120 HU was reached in the ROI placed in the aorta, CT scanning was automatically initiated 5 seconds later. During these 5 seconds, a breath-hold command was given to the patient. Data acquisition was performed in the craniocaudal direction by using 0.75-mm-thick sections, a table feed of 11 mm per rotation, and a 0.5-second gantry rotation period. The tube voltage setting was 120 kV, and the tube current was 200 mAs. All abdominal scans were obtained during breath holds.
For study purposes, three sets of transverse images of each lower leg were reconstructed by using a field of view of 20 cm (ie, targeted reconstruction for the right and left lower legs separately) and different section widths and reconstruction intervals. CT data set 1 had a section width of 2.0 mm and a reconstruction interval of 1.0 mm, CT data set 2 had a section width of 1.0 mm and a reconstruction interval of 0.5 mm, and CT data set 3 had a section width of 0.75 mm and a reconstruction interval of 0.4 mm.
A whole-body patient fixation system (BodyFix; Medical Intelligence, Schwabmuenchen, Germany) was used in all CT angiography examinations to minimize motion artifacts stemming from movement of the body and legs (10).
Three-dimensional CT Reconstructions
All six CT data sets per patient (ie, three data sets with different transverse resolutions obtained in each leg) were transferred to a Advantage Windows workstation (GE Medical Systems Europe, Buc, France) for creating volume-rendered images of the vessels of the lower limbs. All reconstructions were performed by the same technician (M.K.), who had 7 years of experience in three-dimensional postprocessing techniques. Display parameters, including width, level, attenuation, and brightness were visually optimized with the first volume rendering and saved as presets for all subsequent vessel reconstructions in the same patient. From each data set, volume-rendered images were created from transverse CT angiography data in 10° steps, resulting in 36 volume-rendered images covering 360°.
Interpretation of CT Angiograms and DSA Images
Image reading was performed in consensus by two experienced cardiovascular radiologists (T.B. and S.W., with 10 and 11 years of experience in cardiovascular radiology, respectively) who were blinded to the clinical data and to the results of the other imaging modality. First, the DSA images were interpreted as hard copies by the two radiologists. One month later, the CT angiograms (volume-rendered and transverse sections) obtained with the three different section widths and reconstruction intervals (2.0 and 1.0 mm, 1.0 and 0.5 mm, and 0.75 and 0.4 mm, respectively) were assessed at an interactive workstation (Advantage Windows 4.0; GE Medical Systems) by the same two radiologists.
CT data sets obtained with different transverse resolutions in different patients were shown to the readers in a random order and in an order that was different from that used at the DSA reading; consecutive reading of different data sets obtained in one patient was avoided. Volume-rendered CT images and the corresponding transverse CT images were shown to the readers together. The runoff vessels of the lower limb (the popliteal artery, tibioperoneal trunk, anterior tibial artery, posterior tibial artery, and peroneal artery) were assessed on the right and the left sides separately; thus, five vessel segments were assessed in each lower limb. Primarily, the volume-rendered images were assessed for the presence of stenoses. In every case of vessel calcification that prevented assessment of the stenosis grade on volume-rendered images, the transverse images were additionally evaluated. When interpreting the transverse images, both readers were allowed to individually adjust window width and level settings for image analysis. The grade and length of the most severe stenosis in every vessel segment (both calcified and noncalcified) were assessed by using the volume-rendered images together with the transverse images. Length measurements were performed with an electronic caliper available on the workstation.
Arterial and Venous Visibility
Arterial visibility—that is, opacification at DSA and attenuation with the three CT angiography techniques—was rated separately for both legs by using a four-point scale on which a score of 1 indicated that visibility was bad and not diagnostic; a score of 2, that visibility was fair, with impaired visibility of small vessels, side branches, and collateral vessels owing to insufficient vessel opacification or attenuation; a score of 3, that visibility was good and assessment was not impaired; and a score of 4, that there was excellent vessel opacification or attenuation of the entire vascular tree. The presence of opacified or well-attenuated veins was graded on volume-rendered CT and DSA images as follows: a grade of 1 indicated that such veins were not present; a grade of 2, that such veins were present but assessment of the arteries was not impaired; and a grade of 3, that assessment of the arteries was impaired owing to the presence of overlying veins.
Stenosis Grade and Length at DSA and CT Angiography
The two readers assessed the most severe stenosis in the popliteal artery, tibioperoneal trunk, anterior tibial artery, posterior tibial artery, and peroneal artery of each leg by using the same scale for DSA images, volume-rendered CT images, and transverse CT images. The degree of stenosis was rated as follows: a grade of 0 indicated the presence of a normal vessel or an irregular vessel with luminal narrowing of less than 10%; a grade of 1, stenosis with luminal narrowing of 10%–49%; a grade of 2, severe stenosis with luminal narrowing of 50%–99%; and a grade of 3, occlusion. When more than one stenotic luminal change was detected in the same vessel segment, the most severe stenosis was used for grading and analysis.
The length of the stenoses in a vessel segment was quantified by using the following scale: a score of 0 indicated no stenosis; a score of 1, stenosis of less than 1 cm in length; a score of 2, stenosis of 1–5 cm in length; a score of 3, stenosis more than 5 cm in length; and a score of 4, multiple stenoses. Stenoses with scores of 1–3 were determined by using the electronic caliper available on the workstation. Measurements were performed on the most severe stenosis in the vessel segment. A score of 4 was further defined as that given when there was more than one consecutive stenosis of more than 50% in diameter in one particular vessel segment. The stenosis length scores given on the basis of the CT angiograms were compared with those given on the basis of the DSA images.
Discordance between DSA and CT Angiography and between CT Angiography Techniques
For every vessel segment with discordant findings between DSA and the three CT angiography techniques, as well as for every vessel segment with discordances in stenosis grade and/or length rating between the three CT angiography reconstruction techniques, the images were reevaluated by one radiologist (T.S., with 5 years of experience) to assess the reason for the discordances.
Contrast-to-Noise Ratio
Contrast-to-noise ratios (CNRs) were calculated for the three CT angiography data sets as follows: For measurement of the arterial attenuation in the popliteal artery (AApop) in Hounsfield units, an ROI was placed by a radiologist (T.S., with 5 of years experience) over the popliteal artery at the level of the joint space of the knee. Depending on the caliber of the popliteal artery, a maximal possible ROI was chosen (mean size, 0.25 cm2 ± 0.07; range, 0.23–0.31 cm2). Image noise (IN)—that is, the standard deviation of the Hounsfield unit measurements within the ROI—was measured in the air adjacent to the knee on the same transverse section by using an ROI of the same size as that used to assess AApop. Measurement of the background signal for determination of the CNR was performed by using a third ROI (AAfat) that was identical in size to the ROI used for the AApop and image noise measurements and was placed in the fatty tissue adjacent to the popliteal artery on the same CT image. CNRs were calculated for every patient and every reconstruction technique with the following equation: CNR = (AApop – AAfat)/IN. Mean CNRs were calculated for the three reconstruction techniques.
Statistical Analysis
Statistical analysis was performed with a commercially available computer program (SPSS 11.5 for Windows; SPSS, Chicago, Ill). The Wilcoxon exact signed rank test was used for the comparison of the arterial and venous visibility rating, the comparison of the stenosis length between DSA and the three CT angiography reconstructions, and the comparison of mean image noise and CNR between the three CT angiography reconstructions. For each patient (n = 17), stenosis length was compared between DSA and the three CT angiography reconstructions, as well as between the three CT angiography reconstructions. The Bonferroni correction was used, so P values of less than .008 were considered to indicate significant differences in arterial and venous visibility ratings and stenosis length ratings between DSA and the three CT angiography reconstructions and between the three CT angiography reconstructions.
For the comparison of mean image noise and CNR between the three CT angiography reconstructions, a Bonferroni-corrected P value of less than .017 indicated significant differences.
Diagnostic accuracy for the detection of stenoses of 50% or greater with the three CT angiography reconstruction techniques, as compared with accuracy with the reference standard, DSA, was assessed as follows: Stenoses were grouped into those of grade 0–1 and those of grade 2–3, and sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy were calculated by using DSA as the reference standard. Positive predictive value was calculated as TP/(TP + FP), where TP is the number of true-positive findings and FP is the number of false-positive findings. Negative predictive value was calculated as TN/(TN + FN), where TN is the number of true-negative findings and FN is the number of false-negative findings. Diagnostic accuracy was calculated as (TP + TN)/N, where N is the total number of tests.
In cases where there were several angiographic projections, the one that showed the most severe stenosis was used as the reference standard. We took into account the clustered nature of the data (ie, the fact that there were not 163 independent vessel segments but instead clusters of segments in 17 patients), so sensitivity and specificity were calculated for each patient and each reconstruction technique individually. The resulting three data sets with 17 sensitivity values and three data sets with 17 specificity values (ie, three reconstructions for 17 patients) were then compared by using the Wilcoxon exact signed rank test to assess the effect of the reconstruction technique on sensitivity and specificity. A Bonferroni correction was performed, and P values of less than .017 were considered to indicate statistically significant differences.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionReferences |
|---|
|
Assessment of the venous visibility ratings revealed that arterial assessment of both legs on volume-rendered CT images was impaired in one patient owing to venous overlay; this impairment was present with all three CT angiography techniques (Table). Image reading in this patient was based solely on transverse images. No problems with venous overlay were encountered with DSA.
Stenosis Grade and Length at DSA and CT Angiography
Sensitivities for the detection of stenoses were 95.6%, 95.6%, and 97.5%, for CT data sets 1, 2, and 3, respectively. These differences were not statistically significant (P > .017). The corresponding specificities were 85.0%, 88.8%, and 95.3%. The differences in specificity were not significant between CT data set 1 and CT data set 2 or between CT data set 2 and CT data set 3 (P > .017). However, the differences in specificity were significant between CT data set 1 and CT data set 3 (P < .017). The accuracy, positive predictive value, and negative predictive value for each of the three CT angiography techniques, as compared with DSA, are listed in the Table.
The stenosis length was rated higher with all three CT angiography reconstructions than with DSA, but these differences were not significant (P > .008). There was no significant difference in terms of stenosis length rating between the three CT angiography reconstructions (P > .008) (Table).
Discordances between DSA and CT Angiography and between CT Angiography Techniques
Compared with ratings at DSA, stenosis grade was rated higher in 16 arterial segments with CT data set 1 (eight grade 1 stenoses were classified as grade 2 stenoses, and eight grade 1 stenoses were classified as occlusions), in 10 segments with CT data set 2 (five grade 1 stenoses were classified as grade 2 stenoses, and five grade 1 stenoses were classified as occlusions), and in six segments with CT data set 3 (three grade 1 stenoses were classified as grade 2, and three grade 1 stenoses were classified as occlusions) (Table). Hence, detection of hemodynamically significant stenoses (>50%) resulted in 16, 10, and six false-positive classifications for CT data sets 1, 2, and 3, respectively. All false-positive classifications regarding hemodynamically significant stenoses (eg, a grade 1 stenosis classified as a grade 2 stenosis and a grade 1 stenosis classified as occlusion) occurred with calcified vessel segments.
Regarding misclassification of grade 2 stenoses as occlusions with CT data sets 1, 2, and 3 (without having influence on sensitivity and specificity), seven of eight, three of five, and two of three segments were calcified, respectively. The remaining segments with grade 2 stenoses that were classified as occlusions were not calcified. In all cases, DSA showed high-grade stenoses, and CT angiography, in retrospect, did not show a patent vessel lumen.
The higher rate of false-positive findings of hemodynamically significant stenoses in CT data sets 1 and 2, as compared with the rate in CT data set 3, was caused by more severe partial volume artifacts that caused "blooming" of the calcified plaques in the former data sets. Figure 1 shows a calcified grade 1 stenosis (ie, 10%–49% stenosis) that was overestimated with CT data sets 1 and 2 as a grade 2 stenosis. The same stenosis was rated correctly as a grade 1 stenosis (<49%) with CT data set 3.
|
|
Rating of stenosis length differed between DSA and CT data sets 1, 2, and 3 for 16, 15, and 14 vessel segments, respectively. The differences between DSA and CT data sets 1, 2, and 3 were larger than one grade for five, six, and six segments, respectively. In these affected segments, grade 1 (<1 cm) and grade 2 (1–5 cm) stenoses were rated as grade 4 stenoses (multiple stenoses > 50% within the segment); this misrating was due to the 360° view of the volume-rendered images depicting more stenoses than DSA. In the remaining 11, nine, and eight vessel segments assigned different ratings at DSA and CT angiography, the length grade differed by one (one, one, and zero segments were underestimated and 10, eight, and eight segments were overestimated with CT data sets 1, 2, and 3, respectively). In the two segments in which length was underestimated with CT data sets 1 and 2, length was close to 1 cm and was classified as grade 2 at DSA and grade 1 at CT angiography. In the 10, eight, and eight imaging studies in which length was overestimated with CT data sets 1, 2, and 3, respectively, the different ratings could be attributed to asymmetrical stenoses that were insufficiently depicted at DSA in seven, six, and seven segments, respectively. In the remaining three, two, and one segments, respectively, the differences were caused by a stenosis length that was very close to 1 or 5 cm.
CNRs
Mean image noise was 12.5 HU ± 2, 13.6 HU ± 1.4, and 15.0 HU ± 2.2 for CT data sets 1, 2, and 3, respectively. The differences were statistically significant between CT data sets 1 and 3, between data sets 1 and 2, and between data sets 2 and 3 (P < .017 for all) (Table).
The CNR of the thinnest reconstruction (CT data set 3) was significantly lower than the CNR of the other two reconstructions (P < .017) (Table). Maximum CNRs were 48.1, 36.4, and 34.5 for CT data sets 1, 2, and 3, respectively. Minimum CNRs were 22.8, 13.5, and 11.7 for CT data sets 1, 2, and 3, respectively.
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionReferences |
|---|
Diagnostic Accuracy of CT Angiography versus DSA
In the present study, sensitivity in the detection of stenoses was excellent (>95%) and did not differ between the three CT angiography techniques. However, the specificity significantly increased when a thinner section width was used. This was due to the fact that a larger number of false-positive stenosis detections occurred with thicker sections. We observed two mechanisms that caused false-positive ratings of hemodynamically significant stenoses. First, false-positive ratings may occur with calcified stenoses that have a stenosis grade of just less than 50% owing to partial volume artifacts that occur with thicker section widths. These artifacts may cause "blooming" of calcified plaques and thus lead to overestimation of stenoses (12). In our study, this mechanism caused the majority of the false-positive ratings at CT angiography. Partial volume artifacts and calcium "blooming" decrease with smaller section widths; therefore, a significantly higher specificity with the reconstruction technique with the smallest section width was observed in this study. Second, shortcomings of the reference-standard technique, DSA, may result in false-positive ratings when stenoses are missed owing to projectional errors. This occurred for two segments in our study.
Occlusions were more frequently diagnosed with CT angiography than with DSA. The distal portions of lower limb vessels are very small in diameter, and a high-spatial-resolution imaging technique is required for detecting such small arteries. Use of the higher z-axis resolution of the thinnest CT angiography reconstruction consequently yielded the lowest false-positive numbers of grade 3 stenoses (occlusions) as compared with the results of DSA.
As compared with previously published results with four–detector row CT angiography of the lower leg (6,13), our results with 16–detector row CT angiography were much better in terms of sensitivity, specificity, and accuracy, at least when CT angiography was performed with the highest z-axis resolution. The rating of stenosis length differed only between DSA and the three CT angiography reconstructions, not between the three CT angiography reconstructions. This overestimation of stenosis length with CT angiography was mainly caused by the use of 360° views with volume-rendered images and by the assessment of transverse CT angiography images, which enable one to detect, in contrast to DSA images, even parts of stenoses that are oriented in a strict anteroposterior orientation. The overestimation with regard to the reference standard DSA can therefore be attributed to shortcomings of this standard and not to the failure of CT angiography to depict the real length of the stenoses.
Results of discordance analysis revealed that in 12 of 16, 12 of 15, and 13 of 14 segments, insufficient depiction of asymmetric stenoses at DSA was the reason for different length ratings with CT data sets 1, 2, and 3, respectively.
Study Limitations
A limitation of our study is related to the low number of patients (n = 17) enrolled. However, taking into account the clustered nature of data, we independently compared the stenosis length and the sensitivity and specificity at DSA with the stenosis length and sensitivity and specificity with the three CT angiography reconstructions for each patient and found significant differences in specificity values between CT data set 1 and CT data set 3. Increasing the number of patients might reveal additional significant differences.
Another limitation is that we performed consensus reading, which is not easily generalizable to typical reading settings.
A third limitation was the focus of the study on lower leg arteries. However, four–detector row CT angiography has already shown excellent concordance with DSA in terms of stenosis grading in the aortoiliac and femoral arteries (9) and stenosis grading of peripheral bypass grafts (14). It seems unlikely that 16–detector row CT, with its higher transverse resolution, might perform in an inferior way in these vascular regions. Taking into account the huge volume of data generated with three reconstructions of the peripheral runoff vessels and the most probably limited gain of information in the aortoiliac and femoral arteries, we focused solely on lower-leg vessels.
Use of the thinnest CT angiography reconstruction and the generation of volume-rendered reconstructions involves some potential disadvantages. Thin transverse images dissipate more archiving resources and slow down network transfer of data sets, and the generation of volume-rendered images may be laborious and time consuming. Archiving only volume-rendered reconstructions is not an option for managing the huge amount of data because transverse images are necessary for the assessment of calcified stenoses. Therefore, archiving the entire data set is inevitable. We use a picture archiving and communication system, or PACS, which is able to handle our huge CT angiography data sets without problems. In the future, all manufacturers of PACS systems will be forced to provide sufficient computing power and bandwidth and a system design that can manage these large amounts of data. Taking into account the significantly better performance of the 0.75-mm section width compared with the performance of thicker reconstructions, the additional expenses for data storage is justified.
The image noise was significantly higher and the CNR was significantly lower with the 0.75-mm CT reconstructions than with the 2.0- and 1.0-mm reconstructions. Nevertheless, this was not obviously visible during assessment of the transverse images and volume-rendered reconstructions, and the diagnostic accuracy of stenosis grading was best with the thinnest section width. The advantage of use of a thin section width, therefore, outweighs the measurable decrease in image quality.
In conclusion, as compared with DSA, 16–detector row CT angiography has excellent diagnostic accuracy in the assessment of lower-leg PAD, provided that the thinnest possible section width is used.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: CNR = contrast-to-noise ratio • DSA = digital subtraction angiography • PAD = peripheral arterial disease • ROI = region of interest
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, T.S., S.W., T.B.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, T.S., T.B.; clinical studies, T.S., S.W., M.K., T.B.; statistical analysis, T.S., T.B.; and manuscript editing, T.S., H.A., T.B.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Thurnher, S. Miller, G. Schneider, C. Ballarati, G. Bongartz, C. U. Herborn, S. Schoenberg, M. A. Cova, G. Morana, K. Niazi, et al. Diagnostic Performance of Gadobenate Dimeglumine Enhanced MR Angiography of the Iliofemoral and Calf Arteries: A Large-Scale Multicenter Trial Am. J. Roentgenol., November 1, 2007; 189(5): 1223 - 1237. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. H. Heijenbrok-Kal, M. C. J. M. Kock, and M. G. M. Hunink Lower Extremity Arterial Disease: Multidetector CT Angiography Meta-Analysis Radiology, November 1, 2007; 245(2): 433 - 439. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
