Abdominal

Imaging

Springer Science+Business Media New York 2014

Published online: 18 January 2014

Abdom Imaging (2014) 39:554561

DOI: 10.1007/s00261-014-0077-1

Venous filling defects on portal venous phase

CT of the abdomen and pelvis: clinical
implications and positive predictive value for
diagnosing deep venous thrombosis
Jeffrey J. Horvath,1 Christopher B. Looney,1 Rendon C. Nelson,2 Charles Y. Kim1
1
2

Division of Vascular and Interventional Radiology, Duke University Medical Center, Box 3808, Durham, NC 27710, USA
Division of Abdominal Imaging, Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC 27710, USA

Abstract
Purpose: To determine the prevalence, resulting clinical
decisions, and the positive predictive value (PPV) of
venous filling defects detected on portal venous phase
(PVP) CT.
Methods: Over a 3-year period, 42412 consecutive
patients underwent a PVP abdominopelvic CT; of these,
348 reports mentioned a filling defect concerning for
deep venous thrombosis (DVT) in the IVC, iliac, or
common femoral veins. Ninety-three patients underwent
a reference standard venous imaging study.
Results: The prevalence of venous filling defects in CT
reports was 0.82% (n = 348). Reports worded with
higher degrees of certainty were statistically more likely
to result in treatment, while lower certainty was correlated with additional confirmatory imaging. The PPV for
detection of DVT was 77%. The presence of peri-vascular
stranding or vessel expansion increased the PPV of PVP
CT to 95% and 100%, respectively.
Conclusion: While the PPV for filling defects on PVP CT
was modest, it was substantially improved if peri-venous
stranding or vessel expansion was present.
Key words: CTPortal venous phaseDeep venous
thrombosis

Lower extremity deep venous thrombosis (DVT) and

resultant pulmonary embolism (PE) can be clinically
occult, resulting in an underestimated true prevalence of
disease, particularly in at-risk populations such as those

Correspondence to: Charles Y. Kim; email: charles.kim@duke.edu

with a history of malignancy. Studies of staging CT scans

in cancer patients have shown a 2.5%7% overall prevalence of asymptomatic pelvic and lower extremity DVT
[1, 2], and unsuspected PE has been reported in 2.6%
3.4% of the same oncologic population [35]. Elsewhere,
occult PE has been reported in up to 1.5% routine CT
scans [6]. At the time of proximal lower extremity DVT
diagnosis, PE is found in up to 50% of patients [7], and
inadequately treated DVT is associated with recurrent
PE [8, 9].
Substantial literature exists regarding the utility and
limitations of dedicated CT venography for detection of
lower extremity or pelvic DVT, with a 2008 meta-analysis of 13 contemporary articles showing sensitivity
ranging from 71% to 100% and specicity ranging from
93% to 100% [10]. CT venography for evaluation of DVT
is typically performed following a CT pulmonary arteriogram with lower extremity and pelvic scanning performed at a delay of 120180 s, which is usually adequate
to allow sufficient enhancement of the lower extremity
and pelvic venous system [11, 12]. However, standard
portal venous phase (PVP) protocols of the abdomen
and pelvis performed to optimize evaluation of the
abdominal and pelvic viscera typically use a 4090 s
delay. While the sensitivity and specificity of dedicated
CT venography of the abdomen and pelvis have been
reported, these values are unlikely to be accurate for PVP
imaging [1315], in which the majority of incidental DVT
is likely detected. Given the phenomenon of venous
mixing artifact and lack of literature on DVT diagnosis
in the PVP, radiologists are forced to qualify or limit
their diagnosis of DVT in the presence of apparent filling
defects on CT, and clinicians may be left with indeterminate interpretations and/or recommendations for
confirmatory studies. Therefore the purpose of this study

555

J. J. Horvath et al.: Venous filling defects on PVP CT

was to determine the incidence of venous filling defects as

detected on PVP CT, the clinical decisions resulting from
this finding, and the specificity of PVP CT for the
diagnosis of DVT.

Materials and methods

Subjects
Institutional Review Board approval and a waiver for
informed consent were obtained for single-center retrospective study. Using a computerized database textsearching program, we identied all CT reports from
January 2007 through December 2009 that included any
mention of a venous lling defect or thrombus. The
search terms used were lling defect, DVT, deep
vein thrombosis, thrombus, thrombosis, and
clot. This initial search produced a list of 4682 studies.
Each report was individually reviewed to identify
abdominopelvic CT scans performed in the PVP with
specic concern for acute or subacute thrombus in the
IVC, iliac veins, or common femoral veins. Reports that
mentioned only arterial thrombus, visceral venous
thrombus, or thrombus in locations other than the
abdomen or pelvis were excluded, as were reports that
specied consequences of chronic venous thrombosis,
such as wall calcication or segmental vessel stenosis.
Thus, a total of 348 patients (170 males, 178 females,
mean age: 57.8 years) underwent PVP CT of the abdomen and pelvis with some degree of concern for DVT.
For this patient population, the PACS database and the
electronic medical records were examined for the presence of a reference standard imaging study of the venous segment of concern within 2 weeks of the CT scan.
The reference standard modalities included lower
extremity Doppler ultrasound, magnetic resonance
venography, and conventional venography, all of which
have been extensively validated in the literature as having
excellent sensitivity and specicity for detection of DVT.
Medical records were reviewed for the acute onset or
resolution of symptoms of DVT, specically that in the
lower extremity. Any patient with a substantial change in
DVT symptoms between the index CT and reference
standard study were excluded to avoid potential inclusion of patients with an interval change in clot burden.
The total number of CT examinations of the abdomen
and pelvis performed in the PVP over the inclusion time
period was determined from the imaging database.

CT portal venous phase technique

Patients undergoing PVP imaging of the abdomen and
pelvis underwent image acquisition on 16 or 64-slice
multidetector CT scanners as the primary contrast-enhanced dataset or after the initial acquisition of arterial
phase images. In all studies, 125150 mL of non-ionic
contrast material (Isovue 300; iopamidol 300 mgI/mL,

Bracco Diagnostics, Inc., Princeton, NJ, USA) was injected at a rate of 35 mL/s through an 1822 gauge
peripheral intravenous catheter or a pre-existing powerinjectable central venous catheter. PVP images in all
patients were obtained at a post-injection time of
approximately 70 s, and images from the lung bases
through the pubic symphysis were generated. The slice
thickness was 5 mm.

Doppler ultrasound technique

Sonographic assessment for DVT included gray-scale
sonography with compression and color Doppler imaging of the common femoral vein, supercial femoral vein,
and popliteal vein of each leg studied. Spectral Doppler
imaging, augmentation, and Valsalva maneuver were
employed as needed. The radiology report was reviewed
for the presence of DVT in the common femoral veins.

MRV technique
Dedicated MR venography was performed on 1.5T
magnet strength scanners (Avanto, Siemens Medical
Systems, Erlangen, Germany or Signa HDx, GE Medical
Systems, Milwaukee, WI, USA). Post-contrast steady
state imaging was performed with contiguous 5 mm axial
sections 5 min after the administration of 0.12 mL/kg
(0.03 mmol/kg) gadofosveset trisodium (Ablavar; Lantheus Medical Imaging, Billerica, MA, USA) followed by
a 2030 mL saline ush through a peripheral upper
extremity intravenous catheter. For the 1.5T Avanto, the
following imaging parameters were used: TR (msec) 5.5;
TE (msec) 2.6; slice thickness (mm) 4; distance between
slices (mm) 0; number of slices 240; acquisition matrix
384 9 307; ip angle 10; bandwidth (hertz/pixel) 305;
parallel imaging acceleration factor: 2. For the 1.5T
Signa, the following imaging parameters were used: TR
(msec) 3.5; TE (msec) 1.7; slice thickness (mm) 61; distance between slices (mm) 0; number of slices 240;
acquisition matrix 288 9 192; ip angle 12; bandwidth
(kHz) 41; parallel imaging acceleration factor: 2 (performed only for the abdomen station). The radiology
report was reviewed for the presence and location of any
DVT.

Conventional venography technique

Conventional venography of the IVC, iliac veins, and
common femoral veins was performed in cases where
catheter-directed thrombolysis or inferior vena cava lter
insertion was performed at the discretion of the primary
medical team. In both procedures, a diagnostic venogram was performed as the initial step in the procedure.
For IVC lter insertion, the patient was placed supine on
the uoroscopy table and venous access was gained via
the right internal jugular vein or one of the common

556

femoral veins. For DVT thrombolysis, the patient was

placed prone on the uoroscopy table and the popliteal
vein was used for venous access. Digital subtraction
angiography was performed with conventional venographic techniques using Isovue 300 contrast material
(iopamidol 300 mgI/mL, Bracco Diagnostics).

Image interpretation and medical record data

collection
All CT reports with mention of venous thrombosis were
analyzed for the perceived level of certainty conveyed by
the report for the diagnosis of DVT. The level of certainty was stratied into levels, per the following 4-tiered
scale (1possible; 2worrisome, concerning; 3probable, likely; 4denite, consistent). The clinical impact
of the PVP CT was determined for patients without a
known diagnosis of DVT from the medical record and
categorized as resulting in treatment (anticoagulation,
thrombolysis, or IVC lter placement), additional diagnostic imaging studies, or non-impactful (no treatment
or symptom monitoring). Finally, patient history of
DVT prior to CT and history of malignancy was determined for each examination.
For patients with reference standard imaging, the CT
images were retrospectively reviewed by a board-certied
radiologist who was blinded to the report ndings. The
location of DVT was recorded by venous segment (iliofemoral, common iliac, IVC, and IVC associated with an
IVC lter). On axial images, attenuation values in the
vessel of concern were measured via an ovoid region of
interest (ROI) that was maximized to t within the lling
defect. When applicable, attenuation values of the contralateral normal vein at the same level, normal vein in a
segment cranial to the abnormality, or normal vein
caudal to the abnormal segment (listed in order of
preference) were recorded. Assessment of the presence or
absence of focal segmental venous expansion was determined based on the reference venous diameters in a
similar fashion, and vein diameter was recorded when a
normal contralateral vein at the same axial level was
available for measurement. The presence or absence of
peri-venous fat stranding was also assessed.

Statistical analysis
Statistical analysis was performed using SAS version 8.2
(SAS Institute, Cary, NC, USA). The association between imaging report language and subsequent action
was assessed using the v2 test or Fishers exact test.
Comparison of positive predictive values (PPVs) based
on the degree of certainty, perivenous stranding, and
luminal expansion was performed using the v2 or Fishers
exact test. The mean attenuation of filling defects on PVP
CT that were proven to represent thrombus on reference
standard imaging were compared to filling defects that

J. J. Horvath et al.: Venous filling defects on PVP CT

were shown to not represent thrombus on reference

standard imaging using the unpaired students t test. For
iliac and common femoral veins, the mean filling defect
attenuation for positive studies was compared to normal vein lumen using the paired t test. Two-tailed p
values computed from Fishers exact test were used to
examine the statistical association between multiple
variables in patients with gold standard imaging within
2 weeks for comparison. A p value of 0.05 or less was
considered to indicate a statistically significant difference.

Results
Over a 3-year period, 42412 abdominopelvic CT scans
were performed in the PVP between January 2007 and
December 2009 at our institution. Review of the radiology reports revealed that 348 had mention of a venous
lling defect suspicious for DVT in the IVC, common
iliac vein, and/or iliofemoral segment, resulting in a
prevalence of 0.82%. In these 348 studies, a lling defect
was reported in the IVC in 107 patients (30%), in the
common iliac vein in 92 patients (26%), and in the iliofemoral segment in 236 patients (68%). Out of the 348
report examinations with suspected DVT, 238 of patients
(68%) had an existing diagnosis of malignancy.

Clinical decision making based on ndings of

venous lling defects on CT report
Of the 348 CT examinations with concern for DVT, 159
(46%) patients had a previous diagnosis of DVT. Of the
remaining 189 patients without a known diagnosis of
DVT, 84 (44%) underwent treatment of DVT (anticoagulation, thrombolysis, or IVC lter placement), 39
(21%) underwent additional imaging for evaluation of
potential DVT, and in 66 (35%), no specic action was
taken.
Based on the language of the radiology report in
patients without a prior diagnosis of DVT, 107 reports
57(%) were graded as high certainty for DVT, 42 (22%)
were of intermediate certainty, and 40 (21%) were of low
certainty (Table 1). Reports with high certainty for DVT
were more likely to undergo treatment for DVT than
reports with low certainty (63% vs. 10%, p < 0.001).
Conversely, reports with low certainty for DVT were
more likely to undergo additional imaging (30% vs. 13%,
Table 1. Clinical decisions made based on mention of a venous filling
defect on CT, stratified by degree of certainty based on report wording,
for patients without previously diagnosed DVT
Additional imaging
Low
Intermediate
High
Total

12
13
14
84

(30%)
(31%)
(13%)
(44%)

Treatment
4
13
67
66

(10%)
(31%)
(63%)
(35%)

No action
24
16
26
39

(60%)
(38%)
(24%)
(21%)

Total
40
42
107
189

557

J. J. Horvath et al.: Venous filling defects on PVP CT

Table 2. PPV based on report certainty

Low
Intermediate
High

TP

FP

PPV

p value

17
19
57

4
14
54

13
5
3

0.24
0.74
0.95

<0.001a
0.007b
0.021c

TP true positive, FP false positive, PPV positive predictive value

a
Low vs. high
b
Intermediate vs. low
c
High vs. intermediate

Table 3. Venous filling defect distribution and PPV based on location

Iliofemoral
Common iliac
IVC

TP

FP

PPV

82
27
22

66
23
18

15
4
4

0.82
0.85
0.82

Comparison of PPVs does not demonstrate statistically significant

differences
TP true positive, FP false positive, PPV positive predictive value, IVC
inferior vena cava

p = 0.027). Sixty percent of reports with low certainty

for DVT resulted in no specific action, compared to 24%
of reports with high certainty (p < 0.001).

Diagnostic accuracy of portal venous phase CT

for diagnosis of DVT
A total of 93 CT examinations with concern for a venous
lling defect had a reference standard imaging modality
for correlation. Of these, a venous lling defect was
identied in the iliofemoral segments in 82 cases, common iliac veins in 27 cases, and IVC in 22 cases, of which
5 were in association with an IVC lter. A total of 72
studies (77%) had conrmed DVT on reference standard
imaging resulting in a PPV of 0.77. The PPV of reports
worded with high, intermediate, and low degrees of certainty were 0.95, 0.74, and 0.24, with each level of certainty resulting in statistically signicant increased PPVs
(Table 2). Analysis of PPV based on vessel location
did not reveal any statistically significant differences
(Table 3).

Ancillary CT ndings to support diagnosis of

DVT
Of the 93 studies with reference standard imaging, 42
(45%) had evidence of vessel expansion on PVP CT
(example Figs. 1, 2, 3). Forty of these studies had confirmed DVT, resulting in a PPV of 0.95. Peri-venous
stranding associated with the venous filling defect on CT
(example Figs. 3, 4) was identified in 27 studies (29%); all
cases had DVT confirmed on correlative imaging,
resulting in a PPV of 1.0. The mean density for filling
defects confirmed to be DVT was 43.4 13.9 HU

Fig. 1. Perivenous stranding on PVP CT associated with a

diagnosis of common femoral vein DVT in a 61-year-old female. A Left common femoral vein (arrow) shows abnormal
stranding in the perivenous fat compared to the right (arrowhead). B DVT was confirmed at ultrasound.

compared to 64.7 21.1 HU for negative studies

(p < 0.001). However, analysis of density cutoff
thresholds of 40, 50, 60, and 70 HU resulted in similar
PPVs (range 0.880.91, p = NS). For filling defects in
the iliac and common femoral veins, analysis of luminal
density contralateral to the filling defect similarly did not
result in differences in the PPV for diagnosis of DVT.
Employing threshold density differences of 20, 40, 60,
and 80 HU, PPV values ranged from 0.86 to 0.98
(p = NS).

Discussion
The vast majority of current literature analyzing the
diagnostic value of CT for DVT focuses on dedicated CT
venography protocols that utilize an extended delay to
maximize contrast enhancement of the deep venous
system of the pelvis and lower extremities. A 2008 metaanalysis comparing dedicated CT venography to gold
standard ultrasound or conventional venography in

558

Fig. 2. Abnormal venous expansion on PVP CT associated

with a diagnosis of common femoral vein DVT in a 91-yearold female. A The left common femoral vein (arrow) shows
abnormal expansion (>30% increase in diameter) relative to
the right common femoral vein (arrowhead). Left common
femoral vein is decreased in attenuation compared to the
right, but streak artifact from hip hardware makes Hounsfield
units unreliable. B Doppler ultrasound imaging shows nearly
occlusive thrombus.

patients with suspected DVT and PE resulted in a pooled

sensitivity and specicity of 96% and 95%, respectively,
for proximal DVT [10]. However, abdominopelvic CT is
very commonly performed in the PVP for optimal
enhancement of solid organs for indications such as
malignancy evaluation and abdominal pain. Standard
PVP protocols are therefore much more common than
CTV in clinical practice and thus likely comprise the vast
majority of examinations in which incidental DVT is
found.
Our study demonstrated an overall low prevalence of
incidental venous lling defects reported on PVP CT of
the abdomen and pelvis. Our IVC, iliac, and common

J. J. Horvath et al.: Venous filling defects on PVP CT

Fig. 3. Both abnormal expansion and peri-venous stranding

in the nonenhancing right common femoral vein (arrow) on
PVP CT in a 52-year-old female. The normal enhancing left
common femoral vein is denoted by an arrowhead. B MR
Venogram true FISP image showing confirmed right common
femoral vein DVT, which demonstrates low intraluminal signal
(arrow) when compared to the normal left common femoral
vein (arrowhead).

femoral vein DVT prevalence of 0.82% is similar to the

0.95% prevalence reported in a prior study of incidentally
identied thromboembolic phenomenon in cancer patients [1]. While similar, our patient population consisted
of a more diverse patient population that was not limited
to patients with cancer, and thus the hypercoagulability
associated with malignancy may account for this minor
difference [16].
Although the diagnostic accuracy of PVP abdominopelvic CT for the diagnosis of DVT has not been
previously validated to our knowledge, our data demonstrate that CT reports of venous lling defects on PVP
CT had a substantial impact on clinical decision making.
For a third of reports stating concern for DVT, treatment for DVT was undertaken without additional correlative imaging. In the subset of reports expressing a
high level of certainty in the diagnosis of DVT, the
majority of patients were treated for DVT. Conversely,
reports expressing a low certainty for DVT resulted in
only 10% being treated and the majority undergoing no
action as a result. Not surprisingly, reports with low to

J. J. Horvath et al.: Venous filling defects on PVP CT

559

Fig. 4. A IVC thrombus (arrowheads) associated with an

IVC filter on PVP CT in an 81-year-old male. B Confirmatory
digital subtraction venography showing nearly occlusive
thrombus in the suprarenal IVC associated with an IVC filter.

intermediate concern for DVT underwent additional

conrmatory imaging more frequently than reports with
high concern. It is uncertain why clinicians sometimes do
not initiate treatment or request additional imaging when
the possibility of DVT is reported and is not feasible to
ascertain on a retrospective review. In the absence of
clinical symptoms such as leg swelling, the clinician may
disregard the concern for DVT. However, asymptomatic
DVTs can still result in PE and thus treatment may be
benecial [17].

Fig. 5. False positive read for DVT in a 52-year-old male. A

Heterogeneous low attenuation filling defect without expansion
or peri-venous stranding in the bilateral right greater than left
femoral veins (straight arrows), read as moderately concerning
for thrombus on PVP CT. B Confirmatory ultrasound was
negative. Apparent filling defect likely due to mixing artifact.

While the overall PPV for detection of DVT on PVP

CT was 77%, there was signicant variability based on
the report wording. Reports mentioning concern for

560

DVT with high degree of certainty, as occurred in the

majority of reports in this study, resulted in a PPV of
95%. This compares favorably with the PPV dedicated
CT venography, which has a PPV of 91% [18]. Additionally, both perivenous stranding and vessel expansion
in association with venous filling defects demonstrated
very high PPVs for the diagnosis of DVT. Venous
luminal expansion and peri-vascular edema have both
been previously described in association with DVT using
ultrasound and MRI, respectively [19, 20]. Although the
mean density of filling defects confirmed to be DVT was
significantly lower than false positive filling defects,
analysis of density thresholds for filling defects did not
improve the PPV. Similarly, comparison of the contralateral and distal venous luminal densities also did not
prove helpful. Several reasons may explain this finding.
Thrombus, while classically defined as a low attenuation
filling defect, can demonstrate varying densities based on
its current stage of evolution and hematocrit level [21].
Furthermore, the degree of venous enhancement in the
lower extremities can be heterogeneous, particularly at
areas of mixing of opacified blood with non-opacified
blood (Fig. 5), and in conditions when there is asymmetric velocity of venous return from the extremities.
This study has several important limitations. Since
the vast minority of patients with a lling defect did not
have reference standard venous imaging, we were not
able to calculate a sensitivity, specicity, or negative
predictive value. Thus, only PPVs were able to be calculated. The retrospective nature of this study limits the
availability of clinical follow-up in this population. Although reference standard imaging was required within
14 days of the index CT scan, thrombus can develop or
resolve in this time period. However, by excluding patients with a change in clinical exam, we hoped to minimize the inclusion of such patients. Another limitation is
the heterogeneity in radiologists rendering the report,
who did not follow any specic guidelines for diagnosing
DVT or conveying their condence level. However, no
guidelines exist for diagnosing DVT on PVP CT.
Therefore, analysis of radiology reports as rendered for
these studies in real life by a diverse group of radiologists may allow our results to be more generalizable and
practical for use by clinicians trying to decide on a course
of action based on the mention of a venous lling defect
on PVP CT.
Our ndings have several important clinical implications. First, while venous lling defects suspicious for
DVT on PVP CT are a relatively rare occurrence, they
occur with enough frequency to be confronted on a
regular basis given the ubiquitous use of PVP CT. Although the diagnostic accuracy of PVP CT for detection
of DVT has not been previously reported, clinicians often act on the radiology report when venous lling defects are mentioned. The PPV was modest at 77%, but in
the setting of condent interpretations, which accounted

J. J. Horvath et al.: Venous filling defects on PVP CT

for the majority of patients in this study, the PPV was

excellent at 95%. While analysis of the lling defect
attenuation did not improve diagnostic accuracy, we
found that the nding of associated peri-vascular
stranding and vessel expansion resulted in PPVs of 95%
of higher. Therefore, peri-vascular standing and/or venous expansion should lead radiologists to produce more
condent diagnosis of DVT, which has the potential to
reduce numbers of conrmatory imaging studies and
improve the rapidity of treatment initiation.

References
1. Douma RA, Kok MG, Verberne LM, Kamphuisen PW, Buller HR
(2010) Incidental venous thromboembolism in cancer patients:
prevalence and consequence. Thromb Res 125:e306309
2. Cronin CG, Lohan DG, Keane M, Roche C, Murphy JM (2007)
Prevalence and significance of asymptomatic venous thromboembolic disease found on oncologic staging CT. Am J Roentgenol
189:162170
3. Gladish GW, Choe DH, Marom EM, et al. (2006) Incidental pulmonary emboli in oncology patients: prevalence, CT evaluation,
and natural history. Radiology 240:246255
4. Storto ML, Di Credico A, Guido F, Larici AR, Bonomo L (2005)
Incidental detection of pulmonary emboli on routine MDCT of the
chest. Am J Roentgenol 184:264267
5. Sebastian AJ, Paddon AJ (2006) Clinically unsuspected pulmonary
embolisman important secondary finding in oncology CT. Clin
Radiol 61:8185
6. Gosselin MV, Rubin GD, Leung AN, Huang J, Rizk NW (1998)
Unsuspected pulmonary embolism: prospective detection on routine helical CT scans. Radiology 208:209215
7. Dorfman GS, Cronan JJ, Tupper TB, et al. (1987) Occult pulmonary embolism: a common occurrence in deep venous thrombosis. Am J Roentgenol 148:263266
8. Hull R, Delmore T, Carter C, et al. (1982) Adjusted subcutaneous
heparin versus warfarin sodium in the long-term treatment of venous thrombosis. N Engl J Med 306:189194
9. Hull R, Hirsh J, Jay R, et al. (1982) Different intensities of oral
anticoagulant therapy in the treatment of proximal-vein thrombosis. N Engl J Med 307:16761681
10. Thomas SM, Goodacre SW, Sampson FC, van Beek EJ (2008)
Diagnostic value of CT for deep vein thrombosis: results of a systematic review and meta-analysis. Clin Radiol 63:299304
11. Yankelevitz DF, Gamsu G, Shah A, et al. (2000) Optimization of
combined CT pulmonary angiography with lower extremity CT
venography. Am J Roentgenol 174:6769
12. Kalva SP, Jagannathan JP, Hahn PF, Wicky ST (2008) Venous
thromboembolism: indirect CT venography during CT pulmonary
angiographyshould the pelvis be imaged? Radiology 246:605611
13. Bruce D, Loud PA, Klippenstein DL, Grossman ZD, Katz DS
(2001) Combined CT venography and pulmonary angiography:
how much venous enhancement is routinely obtained? Am J
Roentgenol 176:12811285
14. Katz DS, Loud PA, Bruce D, et al. (2002) Combined CT venography and pulmonary angiography: a comprehensive review. Radiographics 22:S319
15. Ghaye B, Szapiro D, Willems V, Dondelinger RF (2002) Pitfalls in
CT venography of lower limbs and abdominal veins. Am J
Roentgenol 178:14651471
16. Nachman RL, Silverstein R (1993) Hypercoagulable states. Ann
Intern Med 119:819827
17. Goodacre S, Sampson F, Thomas S, van Beek E, Sutton A (2005)
Systematic review and meta-analysis of the diagnostic accuracy of
ultrasonography for deep vein thrombosis. BMC Med Imaging 5:1
13
18. Baldt MM, Zontsich T, Stumpflen A, et al. (1996) Deep venous
thrombosis of the lower extremity: efficacy of spiral CT venography
compared with conventional venography in diagnosis. Radiology
200:423428

J. J. Horvath et al.: Venous filling defects on PVP CT

19. Hertzberg BS, Kliewer MA, DeLong DM, et al. (1997) Sonographic assessment of lower limb vein diameters: implications for
the diagnosis and characterization of deep venous thrombosis. Am
J Roentgenol 168:12531257
20. Erdman WA, Jayson HT, Redman HC, et al. (1990) Deep venous
thrombosis of extremities: role of MR imaging in the diagnosis.
Radiology 174:425431

561

21. Kirchhof K, Welzel T, Mecke C, Zoubaa S, Sartor K (2003) Differentiation of white, mixed, and red thrombi: value of CT in
estimation of the prognosis of thrombolysis phantom study.
Radiology 228:126130

Copyright of Abdominal Imaging is the property of Springer Science & Business Media B.V.
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.