Pulmonary embolism (PE) is the third most common acute cardiovascular disease after myocardial infarction and stroke, and results in thousands of deaths each year because of the various clinical presentation.1,2 Diagnostic tests for thromboembolic disease include: (1) the D-dimer assay, which has a high sensitivity but poor specificity in this setting,3 (2) ventilation-perfusion scintigraphy, which has a high sensitivity but very poor specificity,4 and (3) lower limb ultrasonography, which has a high specificity but low sensitivity.5 And pulmonary digital subtraction angiography is still performed as the gold standard in diagnosis of PE.
Today multi-detector computed tomography (MDCT) has already found its place in the evaluation of PE and is in many centers the first line investigation method.6 The recent advent of 64-detector row spiral CT, which can be used to examine the entire chest with 1-mm section thickness during one breath hold, further improved the accuracy of spiral CT in the diagnosis of PE. Another consideration is that the short scanning time of multi-detector CT might provide us with an opportunity to use contrast medium more efficiently. With multi-detector CT, the amount of contrast medium might be reduced without affecting the image quality. Our purpose of this prospective study was to compare the quality of images by using two acquisition protocols with multi-detector CT with different amounts of contrast medium injection in CT pulmonary angiography (CTPA) to determine if it is possible to decrease the dose of contrast medium.
From March 2006 to March 2007, 117 (78 men and 39 women) consecutive patients were referred to our institution for management of PE. All patients underwent multi-detector helical CTPA procedures preceded by physical examination. Other diagnostic tests included plasma D-dimer assay (n=106), ventilation-perfusion scintigraphy (n=46) and lower limb ultrasonography (n=25). Sixty-five patients (44 men and 21 women, mean age (63±15) years, range 30 years-89 years) were excluded from the study because of pulmonary conditions, including a history of lung surgery (n=1), interstitial fibrosis (n=8), parenchymal infiltration (n=7), chronic obstructive lung disease and airway infection (n=14), focal bronchiectasis (n=3), mass (n=1), multi-nodules and a solitary nodule close to the pulmonary vessels (n=8), also involving enlarged lymph nodes in the mediastinum or hilums (n=6), pulmonary hypertension (n=6), heart enlargement (n=5), pleural and pericardial effusion (n=13) and embolus in the main pulmonary artery (n=5). The other 52 patients were randomly assigned to two groups and underwent multi-detector CT with different acquisition protocols and contrast medium injection. Of the 52 patients, 2 were excluded for poor image quality and another 10 cases were excluded when pulmonary embolism of segmental and subsegmental arteries were confirmed clinically. A total of 12 examinations were excluded from further analysis, and finally 40 patients (25 male, 15 female, mean age (64±13) years) with a final clinical diagnosis of normal after 6-month follow-up were eligible to enter the study group.
Group A included 18 patients (12 men and 6 women, mean age (67±15) years, range 30 years-89 years), while 22 patient (13 men and 9 women, mean age (62±11) years, range 44 years-87 years) were in group B.
All patients underwent non-enhanced CT scan of the thorax before CTPA. Patients in group A were evaluated with a 16-detector helical CT (Light-speed 16-detector CT GE, America) using 16×1.25 mm collimation and a table feed of 27.5 mm/revolution, a pitch of 1.375, 120 kV, 300 mA, and 0.56 second rotation. On the basis of these data sets, transverse images were reconstructed with an interval of 1 mm by standard algorithm with a matrix of 512×512 pixels. The mean duration of data acquisition was 5 seconds-7 seconds. All studies were performed with 70 ml of 300 mg/ml Iohexol (Omnipaque 300; Amersham Health Ltd. Shanghai, China) administrated at a rate of 4 ml/s with an automatic dual chamber injector (Medrad) through a peripheral 18-gauge venous access needle that was placed in an antecubital vein. After termination of contrast agent administration, 30 ml of saline was injected. Start delay time was determined with a test injection of 10 ml of contrast material at a rate of 4 ml/s. Time-attenuation curves over the right pulmonary artery were produced from images obtained every second for 20 seconds. Start delay was determined by adding 2 seconds to the time-to-peak value. Scanning delays were 8 seconds-14 seconds (mean 12 seconds).
Patients in group B were evaluated with a 64-detector helical CT (Philips Brilliance 64-detector CT, Holland) using 64×0.625 mm collimation and a table feed of 44.3 mm/revolution, a pitch of 1.11, 120 kV, 300 mA, and 0.5 second rotation. On the basis of these data sets, transverse images were reconstructed with an interval of 0.7 mm. The mean duration of data acquisition was 2 seconds-4 seconds. All studies were performed with 20 ml of 370 mg/ml Iopamiro (Iopamiro 370; Bracco Sine Pharmaceutical Corp. Ltd. Shanghai, China) administrated at a rate of 5 ml/s with an automatic dual chamber injector (Medrad). Start delay time was determined with a test injection of 10 ml of contrast material at a rate of 5 ml/s. Start delay was determined by adding 2 seconds to the time-to-peak value. Scanning delays were 7 seconds-9 seconds (mean 8 seconds).
All patients underwent cranio-caudal scanning in a supine position and at end-inspiratory suspension during a single breath hold. The z-axis coverage and the field of view were chosen to include the entire thorax, from the apex to the base of the lungs.
Image reading was performed at a workstation (Extended Brilliance Workspace workstation) at mediastinal (window width, 360 Hu; window center, 60 Hu) window settings. The radiologists could obtain multiplanar reformatted images only once for each artery. Two experienced radiologists (QIU Jian-xing and ZHU Ying) who were blinded to the contrast medium volume read the images independently to identify analyzable segmental and subsegmental arteries and evaluated factors that cause arteries to be nonanalyzable and misdiagnosed. Any discrepancies concerning the presence or absence of artifacts were resolved by consensus.
To identify segmental and subsegmental arteries, we used the nomenclature as outlined by Remy-Jardin et al.7 This nomenclature is based on the standard descriptions by Jackson, Huber8 and Boyden,9 with slight modifications to account for anatomic variations and the orientation of vessels in the transverse plane on CT scans (Table 1). Twenty segmental and forty subsegmental arteries were described in this nomenclature for each person. According to the quality of the image, we scored the artery from 1 to 3. When an artery was found in the expected, prevailing anatomic location and showed definite contrast enhancement without filling defects from its proximal to distal portions, it would get a 3 scores; if arteries were defined as relative hypoattenuated or without sharp margin, they would get a 2 scores; and if arteries were defined with a false filling defect within the vessels’ lumen that lead to misdiagnosis of PE, they would get a 1 score.
To calculate the degree of contrast enhancement of the pulmonary artery, the radiologists measured the mean CT value (in Hounsfield units) of the main pulmonary artery on the unenhanced and enhanced images by using a circular region of interest (ROI) cursor, which was chosen to be half the diameter of the vessel. Contrast enhancement was calculated by subtracting the CT value of unenhanced images from enhanced images.
To analyze factors that caused misdiagnosis of PE, two experienced radiologists (QIU Jian-xing and ZHU Ying) evaluated the kinds of artifacts.
A partial volume artifact was the result of axial imaging of an axially oriented vessel, which was defined as an apparent filling defect without sharp margins but contiguous images would not demonstrate more apparent filling defects.
A flow-related artifact was the transient interruption of contrast enhancement that appeared as ill-defined margins and higher attenuation (above 78 Hu). Bilateral lower lobe flow-related artifacts due to the poor mixture of blood and contrast material could cause misdiagnosis of PE.
Beam-hardening streak artifacts from dense contrast material within the superior vena cava were commonly seen. This artifact was defined by its nonanatomic, poorly defined, radiating nature, which overlie the right pulmonary and upper lobe arteries.
False filling defects might be demonstrated within the pulmonary veins. Generally, arteries coursed adjacent to the corresponding bronchi, with the exception of the apical-posterior segment of the left upper lobe and the lingular arteries, which might course independently for a short distance before rejoining the bronchi. If the pulmonary vein filled with a poor mixture of unenhanced blood and contrast material it would result in misdiagnosis of PE. The contrast enhancement of the pulmonary vein greater than 25 Hu was defined as pulmonary vein enhanced. The mean CT value (in Hounsfield units) of the left atrium was measured by using a circular ROI cursor.
All analyses were performed by using a software program (SPSS for Windows, version 10.0.0). Mean age, weight and sex ratio of patients in the two groups (n=22, 18) were compared by means of a nonpaired Student's t test for independent samples and a χ2 test.
To compare the image quality of two different groups a normality test, independent samples t-test, Mann-Whitney U test and Pearson chi-square test were used. A P value less than 0.05 was considered to indicate a statistically significant difference. The κ coefficients for interobserver agreement of the imaging studies were calculated.10 A κ coefficient less than 0.20 was interpreted as poor, a κ coefficient between 0.21 and 0.40 as fair, a κ coefficient between 0.41 and 0.60 as moderate, a κ coefficient between 0.61 and 0.80 as good, and a κ coefficient between 0.80 and 1.00 as very good.
A range of patient weights in group A was 48 kg-92 kg (mean weight: (66.5±11.2) kg) and in group B was 55 kg-75 kg (mean weight: (65.2±6.1) kg). In the study there were no significant differences in age (P=0.27), weight (P=0.66) and sex (P=0.62) between the two groups.
Totally 800 segmental and 1600 subsegmental arteries were analyzed. The mean score of segmental pulmonary arteries per patient in both groups were 51.9±1.5 for group A, 58.8±2.1 for group B, respectively. No statistically significant differences were found between the total score of segmental arteries (P=0.925) by Mann-Whitney U test. When each segmental artery was considered separately no differences were found for the score of analyzable arteries between the two groups. A score of less than 2.7 was observed for the anterior segmental artery of the right upper lobe (RA2) (group A, 2.16; group B, 2.59; P=0.459). No segmental arteries with a 1 score were found in either group. On the segmental level, very good interobserver agreement was achieved for group A (κ coefficient =0.852) and good agreement for group B (κ=0.773).
For subsegments no significant difference (P=0.661) was found in the total score of arteries between groups A (116.1±3.9) and B (115.8±3.3). For each separate subsegmental artery no statistic difference was found but slight differences above a significant level of 0.05 were recognizable for the lateral subsegmental of both upper lobes (RA2a: group A 2.83; group B 2.32; P=0.135. RA3a: group A 2.83; group B 2.18; P=0.076. LA3a: group A 2.83; group B 2.45; P=0.236. LA2a: group A 2.83; group B 2.45; P=0.236), inferior subsegmental of the medial in the right middle lobe (RA5b: group A 2.33; group B 2.87; P=0.097), inferior lingular lobe (LA5a: group A 2.33; group B 2.87; P=0.097) (LA5b: group A 2.50; group B 2.87; P=0.209), and the medial subsegmental of the anteromedial basal in the left lower lobe (LA7b: group A 2.17; group B 2.73; P=0.127) (Tables 1 and 2)
The vascular zones which scored lower than 2.5 showed a clear pattern. The lateral rami of the anterior and posterior segmental arteries in the right upper lobes (RA2a and RA3a) of group B, and the subsegmental arteries in the middle lobe and lingular lobe (RA5b and LA5a) of group A, and the paracardiac segments in the left lower lobes (LA7b) of group A were considered to be not adequately depicted.
Mean contrast enhancement in the main pulmonary artery of group A was (343.5±73.8) Hu (range from 235.4 Hu to 464.3 Hu), while in group B it was (265.7±42.5) Hu (range from 203.4 Hu to 350 Hu). The enhancement in the main pulmonary artery of group A was significantly higher than that of group B. In group A the contrast enhancement of one case (5.6%) was lower than 250 Hu, while in group B there were eight patients (36.4%) lower than 250 Hu.
Causes of misdiagnosis of PE included beam-hardening streak (Figure 1), flow-related artifacts (Figure 2), partial volume artifacts and pulmonary vein enhancement (Figure 3). Of group A there were two cases of flow-related artifacts (11.1%), 5 cases of beam-hardening streak artifacts (27.8%), 7 partial volume artifacts (38.9%) and 11 cases with pulmonary vein enhancement (61.1%). There were also 8 cases of flow-related artifacts (36.4%), 3 cases of beam-hardening streak artifacts (13.6%), 5 cases of partial volume artifacts (22.7%) and 2 cases with vein enhancement (9.1%) in group B. No statistically significant difference was found for the incidental rate of beam-hardening streak artifacts, flow-related artifacts and partial volume artifacts between the two groups. Of group B there were slightly more cases of flow-related artifacts than group A (P=0.070), but a lower incidental rate of beam-hardening streak (P=0.272) and partial volume artifacts (P=0.273) (Table 2) without significant difference. For group A a significantly higher incidental rate of pulmonary veins enhancement was found than in group B (P=0.001). The interobserver agreement for flow-related artifacts was moderate (κ=0.51), and for beam-hardening streak was good (κ=0.78) and very good for pulmonary vein enhancement (κ=1) (Tables 3 and 4).
Computed tomographic pulmonary angiography has become the standard of care at many institutions for the evaluation of patients with suspected PE.11 The progress of multi-detector CTs and thinner collimation has improved the image quality of segmental and subsegmental arteries on CT scans. As contrast medium is wildly used, more adverse reactions to contrast medium have been found, especially contrast medium-induced nephrotoxicity. The findings of recent reports indicate that the development of contrast medium-induced acute renal failure (ARF) after diagnostic CT angiography is associated with prolonged hospitalization, marked increases in morbidity, and early and late mortality. There are three relatively distinct mechanisms or pathways proposed for the pathophysiology in contrast-induced ARF: hemodynamic effects, direct contrast medium molecule tubular cell toxicity, and endogenous biochemical disturbances such as an increase in oxygen free-radicals and/or a decrease in antioxidant enzyme activity.12 Reducing the amount of contrast medium used during CT scanning could decrease the incidental rate of the adverse reactions including induced ARF, and could also cut the cost of the procedure. Usually 70 ml-120 ml of contrast medium is administrated during an enhanced scanning process. So in our study, we tried to use only 20 ml, 1/3 the amount of the routine dose of 370 mg/ml Iopamiro, with a 64-detector CT scanning for CTPA in one group, and compared the quality of images with routine scanning protocol. In the present study, the two scanning techniques for CTPA with different amounts of contrast medium could show segmental and subsegmental pulmonary arteries clearly. There was no significant difference between the two groups in depicting the analyzable pulmonary arteries.
On the segmental lever, no artery was found nonalyzable in our study, however misdiagnosis of PE in the right upper lobe (R2) was found in both methods caused by beam-hardening streak artifacts, and there was a slight difference between the two groups, with lower levels in group B. Because the amount of medium used in group B was less, and could pass through the superior vena cava more quickly than in group A, the density of contrast within the vena cava during the data acquisition was lower in group B.
Although the number of optimally visualized subsegmental arteries shown in both groups was not significantly different, there were still slight differences for separate subsegmental arteries in the two methods. In our study blurring of the image of the lateral rami of the upper lobe (RA2a, RA3a, LA2a, La3a) was higher in group B. That was also found in the preceding studies and was explained by inadequate contrast enhancement and anatomic variance.7,13,14 There was a significant difference for the contrast enhancement of the main pulmonary artery between the two groups. Although no one had reported the minimum degree of pulmonary artery enhancement that was required for assessing the diagnosis of pulmonary emboli, Bae et al15 thought contrast enhancement of 250 Hu seemed adequate based on their clinical experience. In our study the contrast enhancement in only one case (5.6%) of group A was less than 250 Hu, but in group B there were eight cases (36.4%). And the mean contrast enhancement of group A was significantly higher than that of group B. So the number of analyzable subsegmental arteries in the upper lobe was less in group B.
The reason for the less than adequate visualization of the lingular vessels and paracardiac subsegmental arteries in the left lobe (LA5a, LA5b, LA7b) may be that these vessels are in the proximity of the moving heart. The scanning time of group B was 2 seconds-4 seconds which was shorter than in group A, the effect of cardiac motion artifacts would be less in patients of group B, so the rate of analyzable arteries of LA5a, LA5b and LA7b was slightly higher in group B.
In our study, the frequency of analyzable arteries in RA5b, LA5a and LA5b was relatively lower. We think that the reason for the suboptimal depiction of the middle lobe and lingular vessels were of small size, anatomic variability, and obliquity to the transverse scanning plane.
Factors that cause misdiagnosis of PE include patient related, technical, anatomic, or pathologic factors. Radiologists need to evaluate the image quality of a CTPA study and determine whether a PE is present. Radiologists should identify which pulmonary arteries are indeterminate and whether additional imaging is necessary.
Flow-related artifact due to poor mixture of blood and contrast material could cause transient interruption of contrast enhancement. Transient interruption of contrast enhancement is likely related to inspiration and to unenhanced blood entering the right atrium, right ventricle, and pulmonary arteries from the inferior vena cava just prior to image acquisition.16 The typical appearance of flow-related artifact is its ill-defined margins and higher attenuation level (above 78 Hu).17 The interobserver agreement of the artifact was moderate (κ=0.51), for the sign was not easy to identify. So further imaging could be necessary to exclude thrombus hidden in poorly enhanced vessels. Because the scanning time of group B was 2 minutes-4 minutes shorter than A, the incidence rate in group B was slightly higher than in group A. As CT scanners become faster, delaying initial image acquisition until approximately 5 seconds after inspiration should allow the poor mixture of contrast material to pass through the pulmonary circulation. So breath-holding in advance is necessary.
The nonanatomic, poorly defined, radiating nature of beam-hardening streak artifacts were easy to confirm, the interobserver agreement was good (κ=0.78). The cause of streak artifacts was dense contrast material within the superior vena cava. Group B used only 20 ml of contrast medium at 5 ml/s, the medium would pass through the superior vena cava more quickly than group A, and the density of contrast agent within the vena cava during the data acquisition was lower in group B. So the incidence rate was a little higher in group A.
Partial volume artifact is the result of axial imaging of an axially oriented vessel, and will become less of an issue with routine use of narrow detector widths. In our study, patients of group B were scanned with 0.625 mm×64 collimation, so the cases with partial volume artifacts were fewer than in group A.
The enhanced pulmonary vein could cause misidentification of an artery. False filling defects might be demonstrated within the pulmonary veins, which could mimic PE. If CT scanning time is shorter than pulmonary circulation, contrast medium might not have entered the pulmonary veins during data acquisition. So the number of cases with pulmonary vein enhancement was significantly fewer in group B.
There are certain limitations to this study that need to be mentioned. The study did not include all consecutive patients in whom PE was suspected. This was partially owing to the study design, which included only normal populations. Comparing different scanning techniques in the same patient with documented PE is hampered by ethical concerns. And we could not account for anatomic variants and assumed that their occurrence would be evenly distributed in our study population.
The contrast medium used by two groups was different, Omnipaque 300 for group A and Iopamiro 370 for group B, which could make a difference in the contrast enhancement between the two groups. Because the amount of contrast medium administrated in group B was only 20 ml, we chose contrast with a high concentration so as to yield a high degree of contrast enhancement.
We knew that body weight was the most important patient-related factor that affects the magnitude of contrast enhancement,18,19 but we only compared the patient weights of two groups and had not thought about the effect of patient’s weight on the dose of contrast agent.
There are many causes of misdiagnosis that include patient-related factors (respiratory motion artifact, image noise, pulmonary artery catheter, flow-related artifact), technical factors (window settings, streak artifact, lung algorithm artifact, partial volume artifact, stair step artifact), anatomic factors (partial volume averaging effect in lymph nodes, vascular bifurcation, misidentification of veins), and pathologic factors (mucus plug, perivascular edema, localized increase in vascular resistance, pulmonary artery stump in situ thrombosis, primary pulmonary artery sarcoma, tumor emboli).11 In our study only 4 factors were involved. Patients with pulmonary conditions which could affect the evaluation of the peripheral vessels were excluded from our study and the pathologic factors were not an issue. The thinner width of the detector and review of sagittal and coronal reformatted images could help to identify the anatomic factors. And for other factors such as image noise, pulmonary artery catheter and window settings, the lung algorithm had little relationship with the scanning techniques.
As 64-detector helical CT has gained widespread acceptance in diagnosis of PE, a small amount of contrast medium and short scanning time with high image quality are needed. In our study, there was no difference in the number of analyzable segmental and subsegmental arteries between the two groups. However, the reason for the nonalyzable items was slightly different, the lower dose of contrast medium (group B) led to less enhancement of the pulmonary artery, which made imaging of arteries of the lateral rami in the upper lobe less adequate, while high temporal resolution reduced cardiac motion artifacts, which could make paracardiac arteries better depicted. Factors of misdiagnosis of PE were also different between the two methods. A small amount of contrast and short scanning time could decrease the incidental rate of beam-hardening steak and pulmonary vein enhancement, but increase the rate of flow-related artifacts.
In conclusion, thin collimation with 20 ml of high concentration contrast agent injection and high temporal resolution can depict the segmental and subsegmental pulmonary arteries clearly as a routine protocol.
1. Anderson FA Jr, Wheeler HB, Goldberg RJ, Hosmer DW, Patwardhan NA, Jovanovic B, et al. A population-based perspective of the hospital incidence and case fatality rates of deep vein thrombosis and pulmonary embolism: the Worcester DVT study. Arch Intern Med 1991; 151: 933-938.
2. Giuntini C, Ricco GD, Marini C, Melillo E, Palla A. Pulmonary embolism: epidemiology. Chest 1995; 107(1 Suppl): 38-98.
3. Brown MD, Rowe BH, Reeves MJ, Birmingham JM, Goldhaber SZ. The accuracy of the enzyme-linked immunosorbent assay D-dimer test in the diagnosis of pulmonary embolism: a meta-analysis. Ann Emerg Med 2002; 40: 133-144.
4. Value of ventilation/perfusion scan in acute pulmonary embolism: results of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED)—the PIOPED investigators. JAMA 1990; 263: 2753-2759.
5. Borris LC, Christiansen HM, Lassen MR, Olsen AD, Schutt P. Comparison of real-time B-mode ultrasonography and bilateral ascending phlebography for detection of postoperative deep vein thrombosis following elective hip surgery: the Venous Thrombosis Group. Thromb Haemost 1989; 61: 363-365.
6. Rathbun SW, Raskob GE, Whitsett TL. Sensitivity and specificity of helical computed tomography in the diagnosis of pulmonary embolism: a systematic review. Ann Intern Med 2000; 132: 227-232.
7. Remy-Jardin M, Remy J, Artaud D, Deschildre F, Duhamel A. Peripheral pulmonary arteries
: optimization of the spiral CT acquisition protocol. Radiology 1997; 204: 157-163.
8. Jackson CL, Huber JF. Correlated applied anatomy of the bronchial tree and lungs with a system of nomenclature. Dis Chest 1943; 9: 319-326.
9. Boyden EA, ed. Segmental anatomy of the lungs. New York, NY: McGraw-Hill; 1955: 23-32.
10. Altman DG, ed. Practical statistics for medical research London, England: Chapman & Hall; 1991: 299-272.
11. Wittram C, Maher MM, Yoo AJ, Kalra MK, Shepard JA, McLoud TC. CT angiography of pulmonary embolism: diagnostic criteria and causes of misdiagnosis. Radio Graphics 2004; 24: 1219-1238.
12. Katzberg RW. Contrast medium-induced nephrotoxicity: which pathway? Radiology 2005; 235: 752-755.
13. Teigen CL, Maus TP, Sheedy PF II, Stanson AW, Johnson CM, Breen JF, et al. Pulmonary embolism: diagnosis with contrast-enhanced electron beam CT and comparison with pulmonary angiography. Radiology 1995; 194: 313-319.
14. Schoepf UJ, Helmberger T, Holzknecht N, Kang DS, Bruening RD, Aydemir S, et al. Segmental and subsegmental pulmonary arteries
: evaluation with electron-beam versus spiral CT. Radiology 2000; 214: 433-439.
15. Bae KT, Tao C, Gürel S, Hong C, Zhu F, Gebke TA, et al. Effect of patient weight and scanning duration on contrast enhancement during pulmonary multidetector CT Angiography. Radiology 2007; 242: 582-589.
16. Gosselin MV, Rassner UA, Thieszen SL, Phillips J, Oki A. Contrast dynamics during CT pulmonary angiogram: analysis of an inspiration associated artifact. J Thorac Imaging 2004; 19: 1-7.
17. Wittram C, Maher MM, Halpern E, Shepard JA. The Hounsfield unit values of acute and chronic pulmonary emboli. Radiology 2005; 235: 1050-1054.
18. Schueller-Weidekamm C, Schaefer-Prokop CM, Weber M, Herold CJ, Prokop M. CT angiography of pulmonary arteries
to detect pulmonary embolism: improvement of vascular enhancement with low kilovoltage settings. Radiology 2006; 241: 899-907.
19. Heiken JP, Brink JA, McClennan BL, Sagel SS, Crowe TM, Gaines MV. Dynamic incremental CT: effect of volume and concentration of contrast material and patient weight on hepatic enhancement. Radiology 1995; 195: 353-357.
Keywords:© 2009 Chinese Medical Association
computed tomographic pulmonary angiography; pulmonary embolism;; pulmonary arteries; contrast media