Dynamic half-Fourier single-shot turbo spin echo for assessment of deep
venous thrombosis: initial observations
Ivan Pedrosaa,⁎, Long Ngob, Jesse Weia, Michael Schustera, Houman Mahallatic,
Martin Smitha, Neil M. Rofskya
aDepartment of Radiology, Beth Israel Deaconess Medical Center, Boston, MA 02215, USA
bDepartment of Medicine, Beth Israel Deaconess Medical Center, Brookline, MA 02446, USA
cDepartment of Radiology, Foothills Hospital, Calgary, Alberta, Canada T2N 2T9
Received 19 June 2008; revised 26 September 2008; accepted 1 October 2008
Objective: The objective of this study was to retrospectively analyze the value of dynamic half-Fourier single-shot turbo spin echo (HASTE)
imaging in patients with suspected deep venous thrombosis (DVT).
Materials and Methods: Fifty-five veins in 24 patients were interrogated using a HASTE sequence with the patients relaxed and in various
degrees of Valsalva. Veins were analyzed for changes in caliber (+CAL) and signal intensity (+SI) or in their absence (−CAL and −SI,
respectively) and compared with the presence of thrombus on gadolinium-enhanced magnetic resonance imaging.
Results: There was no thrombus in veins with the +CAL, +SI pattern (n=40) (Pb.01). Five of seven veins (71.4%) with the −CAL, −SI
pattern had thrombus (Pb.01). A qualitative change in CAL had a sensitivity of 100% and a specificity of 91% for the presence of thrombus.
An increase of 1.5 mm in CAL had a sensitivity of 100% and a specificity of 93% for this diagnosis.
Conclusion: Dynamic HASTE imaging offers a physiological method to evaluate veins for deep venous thrombosis.
© 2009 Elsevier Inc. All rights reserved.
Keywords: Magnetic resonance imaging; Deep venous thrombosis; Venography
Deep venous thrombosis (DVT) is a common disorder,
with as many as 600,000 patients afflicted annually . DVT
can occur as a complication of medical and surgical
procedures , particularly following orthopedic surgery
, and is seen with increased incidence in patients with
malignant neoplasms .
The morbidity associated with DVT is substantial and,
along with pulmonary embolism, includes acute pain,
swelling of the affected limb and postphlebitic syndrome
from valve damage. These associated illnesses, along with
DVT-associated mortalities, create the imperative for rapid
diagnosis and initiation of appropriate treatment [2–4].
DVT, typically employing color Doppler methods and
dynamic techniques, with the latter including compression
and Valsalva maneuvers [5–8]. These techniques have high
sensitivities and specificities for symptomatic thrombi in the
common femoral and superficial femoral venous systems
[9,10]. An increase in the diameter of the vein as determined
with US during a Valsalva maneuver followed by a prompt
return to the original configuration isassociated with absence
of thrombus in the iliofemoral venous system .
Magnetic resonance (MR) contrast-enhanced venography
has emerged as a valuable imaging tool for the diagnosis of
DVT of the central veins. While initial strategies were based
on unenhanced techniques such as two-dimensional time-of-
flight (TOF) and phase-contrast sequences, subsequent
studies have reported superior accuracy and efficacy with
gadolinium-enhanced MR venography (MRV) [11–14].
Half-Fourier single-shot turbo spin echo (HASTE) is a
technique that provides rapid T2-weighted images that are
relatively immune to motion artifacts. It has been used for a
variety of body-imaging applications [15,16] and more
Available online at www.sciencedirect.com
Magnetic Resonance Imaging 27 (2009) 617–624
⁎Corresponding author. Tel.: +1 617 6679568; fax: +1 617 6677917.
E-mail address: email@example.com (I. Pedrosa).
0730-725X/$ – see front matter © 2009 Elsevier Inc. All rights reserved.
recently for dynamic imaging, including evaluation of pelvic
floor insufficiency . For the latter application, we have
observed venous brightening as well as a caliber (CAL)
change in patients without suspected venous pathology as a
response to straining procedures. The increase in signal
intensity (SI) represents a distinct change compared with the
typical signal void of flowing blood (an exit phenomenon).
The strain-induced venous brightening likely emanates
from a temporary stagnation of venous flow such that the
inherently long T2 relaxation time of blood can be captured.
Such a finding is clinically appreciated with slow flow or
pooling of blood; it is typified by the bright signal
characteristically seen within hemangiomas on T2-weighted
images [18,19]. Similar to US examinations using Valsalva
maneuvers, we have used dynamic HASTE with Valsalva
maneuvers in patients with suspected venous thrombosis as a
rapid physiological adjunct to MRVexaminations prior to the
administration of gadolinium. The purpose of this work was
to retrospectively evaluate the diagnostic value of dynamic
HASTE imaging in the diagnostic workup of patients with
2. Materials and methods
2.1. Study subjects
in 24 patients (10 males and 14 females) referred for MRV
were interrogated with the dynamic HASTE technique.
MRV protocol and used liberally for rapid assessment of
patients with suspected venous thrombus at the discretion of
the monitoring radiologist. The clinical indications for MRV
were as follows: search for pulmonary embolism source
(3 patients), follow-up of known thrombus (4 patients),
extremity swelling (16 patients) and fever of unknown origin
(1 patient). Patients ranged in age from 25 to 80 years, with a
mean of 48 years. This retrospective review was approved by
the IRB of our institution, and informed consent was waived.
2.2. MR imaging
MR examinations were performed with one of our
clinical 1.5-T MR scanners (Vision, Symphony, Quantum;
Siemens, Erlangen, Germany) with a four-channel body
phased-array coil, and anatomic coverage was tailored to the
Prior to administration of contrast, dynamic HASTE
sequences were performed. Parameters were as follows:
TR/TE (effective)/flip angle (1100 ms/64 ms/130°); thick-
ness=5 mm; matrix=204×256; FOV=300–450; and rectan-
gular FOV tailored to patient body habitus. One imaging
slice was prescribed perpendicular to the long axis of the
vein being evaluated such that the vein is visualized in its
short axis. A total of seven images was acquired in the
prescribed anatomic location, with each image acquired
independently in a sequential fashion.
The first image was acquired with the patient in end
expiration and relaxed. After a 7-s break between breathing
commands (i.e., “breathe in, breathe out”), the patients were
asked to hold their breath in end expiration while performing
a Valsalva maneuver, the latter to the command of “bear
down.” Five sequential Valsalva procedures were performed
in mild, moderate, severe, moderate and mild degrees of
straining, followed by a final relaxed position. Between each
effort, patients were asked to relax and instructions for the
subsequent strain procedure were provided. Each image was
acquired in approximately 1 s, with 7 s between subsequent
efforts, for a total dynamic HASTE scan time of 35 s.
Gadolinium-enhanced MRVwas performed using a three-
dimensional (3D) fat-suppressed T1-weighted gradient-echo
sequence with the following parameters: TR=4.2–4.5 ms;
TE=1.9–2.2 ms; flip angle=25°; matrix=160×256–320; and
slice thickness=3–5 mm before interpolation. The slice
orientation for the volumetric acquisition was established on
the basis of required anatomic coverage. In all patients, 3D
T1-weighted gradient-echo images were obtained before and
after the administration of a double dose (0.2 mmol/kg of
body weight) of gadopentetate dimeglumine (Magnevist,
Contrast material was administered with a power injector
(Spectris, Medrad, Indianola, PA, USA) in a biphasic manner
(0.1 mmol/kg of body weight at 2 ml/s immediately followed
by 0.1 mmol/kg of body weight at 0.8 ml/s). This was
followed immediately with a 20-cc saline flush.
Three postcontrast breath-hold sequences were per-
formed, with the first timed for the arterial phase. The time
delay for the arterial phase was calculated using a timing
examination of 1 ml of gadopentetate dimeglumine followed
by 20 ml of normal saline solution at a rate of 2 ml/s as
previously described . The second and third acquisitions
were performed at 40 and 90 s after the arterial phase,
respectively. Subtraction (second and third postcontrast
acquisitions minus the arterial phase, respectively) imaging
was performed in order to enhance the conspicuity of the
venous structures .
2.3. Qualitative image analysis
The HASTE images were retrospectively reviewed by
two fellowship-trained authors with 2 years (JW) and 1 year
(MS) of experience in MR imaging; results were given by
consensus. A third fellowship-trained radiologist (IP) with 6
years of experience served as a referee for cases of
disagreement. The images were evaluated without knowl-
edge of the patients' clinical history and other imaging data.
The dynamic HASTE images were analyzed for sub-
jective change in CAL and/or SI within the venous segment
being evaluated. Progressive enlargement of the evaluated
venous segment with increasing Valsalva efforts was
interpreted as a positive change in CAL (+CAL). Similarly,
a subjective increase in SI (black to gray or white) within the
lumen of the vessel with increasing Valsalva efforts was
interpreted as a positive change in SI (+SI). The absence of
618I. Pedrosa et al. / Magnetic Resonance Imaging 27 (2009) 617–624
such findings was interpreted as a negative change in CAL
(−CAL) and that in SI (−SI). The results for each venous
segment analyzed were tabulated independently and com-
pared with the results of the gadolinium-enhanced MR
2.4. Quantitative image analysis
One reviewer (JW), who was blinded to the clinical and
imaging findings on contrast-enhanced MR images, mea-
the HASTE images. The maximum and minimum diameters
of the venous segments during the Valsalva maneuvers were
measured in millimeters using electronic calipers on a PACS
workstation (Centricity 2.0, GE Healthcare, Waukesha, WI,
HASTE imaging were calculated by subtracting the mini-
mum CAL from the maximum CAL.
The maximum and minimal SIs within the venous
segment were measured with an ellipsoid region of interest
that encircled the maximum possible area of the vein lumen
without including the wall of the vessel. Changes in SI are
reported as the percentage of differences in SI in the venous
segment interrogated with dynamic HASTE. Percentage
differences were calculated using the formula (SImax−SImin)/
SImin, where SImaxand SIminare the maximum and minimum
SIs during the Valsalva effort, respectively.
2.5. Proof of diagnosis
Each patient's chart and any subsequent imaging test
were reviewed. Proof of diagnosis was based on the
gadolinium-enhanced MR images along with correlations
to the physical findings and other imaging studies. Filling
defects within the vein located centrally (i.e., toward the
heart) with respect to the imaging plane of dynamic HASTE
imaging were considered positive for thrombus. Absence of
filling defects or venous thrombus located peripherally (i.e.,
away from the heart) relative to the dynamic HASTE
imaging plane was considered as negative for thrombus.
Use of gadolinium-enhanced MR images as a gold
standard deserves comment. The reported sensitivity and
specificity of gadolinium-enhanced MRV for the diagnosis
of venous thrombosis are both 100% [12,13]. While these
data support its use as a reference standard, we also pursued
correlative imaging and clinical follow-up for further
confirmation in all cases, with details in Section 3.
2.6. Statistical analyses
The number of veins evaluated per patient ranged from
1 to 4. The within-patient veins induced within-subject
correlation, which needed to be taken into account in all our
analyses. We used a linear mixed-effects model  to
compare the quantitative change in CAL and signal data
between the presence and absence of change assessed by a
qualitative method. In this linear mixed-effects model, we
used compound symmetry to model the correlation of the
within-subject veins. We estimated the sensitivity and
specificity of the qualitative method where the gold
standard of assessment was venous thrombus. Since we
had repeated-measures data, we first computed these
estimates for each person and then took the weighted
estimate from all persons. The algorithm for computing
sensitivity and specificity for repeated-measures data was
proposed by Zhou et al. . The sensitivity and specificity
analysis was done for CAL change, SI change and the
combined CAL and SI change. The optimal threshold of
quantitative CAL change and SI change that would give the
highest diagnostic accuracy of venous thrombus was
interrogated using a receiver operating characteristic
(ROC) curve analysis.
All statistical analyses were performed using Statistical
Analysis System (SAS) version 9.1 (SAS Institute, Cary,
Fifty-four veins were independently analyzed in
24 patients, including the superior vena cava (n=1); right
(n=1) and left (n=1) internal jugular veins; right (n=4) and
left (n=3) subclavian veins; left (n=1) brachiocephalic vein;
inferior vena cava (n=3); right (n=6) and left (n=6) iliac
veins; right (n=12) and left (n=12) femoral veins; and right
(n=2) and left (n=2) popliteal veins. Seven of the 24 patients
(29%) had venous thrombosis. Thrombus was present in 8 of
the 54 veins (15%) evaluated and was located at the level of
(n=5) or central to (n=3) the dynamic HASTE imaging plane.
Two patients had venous thrombosis distal to the HASTE
imaging plane and were considered negative for venous
thrombosis for the purposes of this study.
Conventional venography was performed in 4 patients for
confirmation of diagnosis and endovascular treatment.
Follow-up MR imaging was obtained for 3 patients. CT
was performed in 1 patient, and follow-up ultrasound was
obtained for 2 patients. Surgical follow-up was conducted on
2 patients. In one of these patients, follow-up MR imaging
was also performed, which showed a spermatic cord lipoma
causing external compression upon a patent vein. Clinical
follow-up was conducted on the remaining 9 patients. There
was no case in which subsequent clinical or imaging findings
were discordant with the MR contrast-enhanced venography
findings. Follow-up imaging or clinical data were not
available for 4 of the 24 patients.
3.1. Qualitative image analysis
The different combinations of CAL and SI are presented
in Table 1. The combination of −CAL, −SI was present in 7
(13%) veins and thrombus was present in 5 (71.4%) of them,
at the level of or central to the imaging slice. The +CAL, +SI
pattern was present in 40 (74.1%) veins, and none of them
had associated thrombus at the level of or central to the
619I. Pedrosa et al. / Magnetic Resonance Imaging 27 (2009) 617–624
Twelve veins (22.2%) demonstrated −CAL with Valsalva
maneuvers at qualitative analysis. Thrombus was present at
the level of or central to the imaging slice in 8 (66.7%) of
these, while no thrombus was present in the other 4 (33.3%).
A +CAL was identified in 42veins (77.8%) without
thrombus at the level of or central to the imaging slice.
Venous thrombus was present in 5 (55.6%) and absent in
4 (44.4%) of the 9 veins (16.7%) that showed no change in
signal (−SI) during Valsalva maneuvers. Thrombus was
present in only 3 (6.7%) of the 45 veins (83.3%) that
The sensitivity and specificity for the diagnosis of venous
thrombus for the qualitative presence of CAL change, SI
change and the combination of the two are presented in
Table 2 (Figs. 1 and 2).
3.2. Quantitative image analysis
The mean and median changes in diameter and SI at
quantitative analysis and their relation with the qualitative
evaluation of these two imaging findings are presented in
Table 3. A subjective (qualitative) change in diameter
correlated with a statistically significant difference in
diameter (in mm) at quantitative analysis (P=.036).
The relations between the quantitative changes in
diameter and SI and the presence or absence of thrombus
are shown in Figs. 3 and 4, respectively. The ROC curve
analysis revealed a sensitivity of 100% and a specificity of
93% [95% confidence interval (CI)=66%–100%] for the
diagnosis of venous thrombus when a change in CAL of
1.5 mm was selected (Fig. 5). The sensitivity and specificity
for the diagnosis of venous thrombus with a 113% change in
SI during Valsalva maneuver were 85% (95% CI=64%–
100%) and 76% (95% CI=60%–92%), respectively (Fig. 6).
Gadolinium-enhanced MRV has demonstrated high
sensitivity, specificity and accuracy in the diagnosis of
venous thrombosis of the central veins of the chest, abdomen
and pelvis. The reported sensitivity and specificity of
gadolinium-enhanced MRV for the detection of DVT are
both 100% [11–13].
However, a rapid technique that does not require
intravenous contrast, such as dynamic HASTE, could be
Individual results in 24 patients for the 54 veins evaluated comparing
qualitative CAL and SI changes using a dynamic HASTE technique relative
to the presence of thrombus as determined by gadolinium-enhanced
Pattern No. of veins Thrombus +Thrombus −
Sensitivity and specificity for CAL change, SI change and the combination
of the two on qualitative analysis
CAL change SI changeCAL change and
aSensitivity is defined as the conditional probability of having both
CAL change and SI change given that there is no thrombus. Specificity is as
the conditional probability of not having change in both CAL and SI given
that there is thrombus. These diagnostic accuracy estimates were computed
taking into account the within-subject correlations.
Fig. 1. Images for a 25-year-old woman presenting with pulmonary embolus
without source. Axial HASTE images (TR=1100 ms, TE=64 ms, flip
angle=130°, thickness=5 mm, matrix=204×256) were obtained at relaxed
state (A) and moderate Valsalva maneuver (B). The normal femoral veins
(arrows) show changes in CAL and SI comparing the relaxed state (A) with
the moderateValsalvamaneuver(B). Gadolinium-enhancedMRimages(not
shown) showed no evidence of thrombus.
620I. Pedrosa et al. / Magnetic Resonance Imaging 27 (2009) 617–624
very useful in certain clinical circumstances. Peripheral
venous access is not always available. Some patients may be
unable or unwilling to remain in the magnet for the amount
of time needed to perform gadolinium-enhanced venogra-
phy. Also, such a technique could be used in circumstances
where the use of contrast poses a relative contraindication,
such as pregnancy and severe renal insufficiency. TOF is
another option in these circumstances for a limited and rapid
assessment or a comprehensive assessment [24,25]. How-
ever, comprehensive TOF approaches can be time-consum-
ing, and the approach, in general, is vulnerable to artifacts
that may mimic thrombus .
For example, in-plane saturation occurs if the acquisition
is not performed orthogonal to the direction of flow .
Also, turbulent flow can produce areas of low SI that can be
misinterpreted as thrombi . A recent report has also
described limitations with another contrast medium-inde-
pendent approach to venography using true fast imaging
with steady-state precession sequence in the evaluation for
The HASTE sequence is a routine strategy in many body-
imaging protocols since it provides rapid T2-weighted
images that are relatively motion insensitive. As a fast
spin-echo technique, it is inherently prone to yielding images
with black blood due to the “exit phenomenon,” in which
signal void results when the excited spins of flowing blood
exit the slice prior to capturing their signal.
Relation between qualitative changes in CAL and SI and quantitative
Change at qualitative analysisP
CAL change (mm)
No. of veins
SI change (mm)
No. of veins
aLinear mixed-effects model with compound symmetry structure for
the var–cov matrix of the error term.
flip angle=130°, thickness=5 mm, matrix=204×256) in the relaxed state
(A) and in extreme Valsalva (B) showing no change in SI or increase in CAL
in the thrombosed subclavian vein (white line in A and B), which is
compressed by the axillary lipoma (black line). (C) Coronal maximum
intensity projection from a subtracted (venous minus arterial) 3D T1-weighted
gradient-echo acquisition (TR=4.2 ms, TE=1.9 ms, flip angle=12°,
thickness=4.4 mm, matrix=160×256) demonstrating multiple venous collat-
erals (white arrows) in the expected location of the right subclavian vein,
which is thrombosed. Multiple venous collaterals are also present in the right
shoulder area (open arrows). Note the normal left subclavian vein (arrow-
heads) for comparison.
621 I. Pedrosa et al. / Magnetic Resonance Imaging 27 (2009) 617–624
Based on our prior observations using HASTE for
assessing pelvic floor instability, an increase in CAL and SI
in response to a Valsalva maneuver typified veins that were,
well-known phenomenon in venous sonography [6–8].
Accordingly, we have applied Valsalva maneuvers in our
clinical practice for rapid MR assessment of the vein of
interest. This retrospective review confirms its utility and
We suspect that the increase in the central venous pressure
that follows from a Valsalva maneuver translates pressure
through a widely patent venous system, explaining the
increased CAL observed in veins without DVT. Our study
suggests that CAL change is an excellent predictor of
occlusive venous thrombus with a sensitivity of 100% and a
specificity of 93% when a CAL change greater than 1.5 mm
between baseline and post-Valsalva maneuver is used. None
of the 42 venous segments evaluated in our series that
demonstrated an increase in CAL (+CAL) with Valsalva
maneuvers had thrombus on gadolinium-enhanced MRV at
the level of or proximal to the imaging slice.
Furthermore, the Valsalva-induced increase in intrathor-
acic pressure can slow the rate of venous flow, minimizing
or eliminating the normally seen flow void. The long T2 of
nonflowing intraluminal venous at 1.5 T has been
previously demonstrated during tourniquet occlusion .
However, the T2 of stagnant blood may also be influenced
by an individual's hematocrit .
We hypothesized that veins with thrombus at the level of
or central to the HASTE imaging slice would be fixed in
CAL regardless of the translated pressure and would not
show an increase in SI. However, the sensitivity of
qualitative change in SI was limited for the diagnosis of
occlusive venous thrombus; lack of change in SI was present
in veins with and those without thrombus. Furthermore, the
optimal quantitative threshold of SI change of 113%
provided only moderate sensitivity (85%) and specificity
(76%) for the diagnosis of occlusive thrombus.
Lack of venous distension on US in response to Valsalva
maneuvers indicates the presence of a proximal occlusion
. Distension of the common femoral vein during
Valsalva maneuver on US is highly suggestive of absence
of occlusive thrombus in the ipsilateral iliac vein .
However, distension of the femoral veinon US inresponse to
Valsalva maneuvers can be seen if the thrombus in the
ipsilateral iliac vein is nonocclusive . None of our
patients had upstream (in the abdomen) or downstream (in
the chest) venous thrombosis when a +CAL pattern was
found on dynamic HASTE imaging. However, based on the
reported experience with US, we anticipate that upstream (in
the abdomen) or downstream (in the chest) nonocclusive
thrombus may be encountered in patients with a +CAL
pattern on dynamic HASTE imaging.
In our series, for the qualitative assessment, the addition
of the SI change to the CAL change did not improve the
diagnostic accuracy of dynamic HASTE imaging. In fact,
sensitivity decreased substantially (from 100% to 63%)
when both qualitative CAL and SI changes were considered,
primarily due to the lack of change in SI in nonthrombosed
veins. Our findings match those of Doppler US studies where
changes in velocity of venous blood flow after Valsalva
maneuver followed by expiration only occurred in 68% of
normal veins .
Fig. 4. Box-and-whisker plot showing the percentage of change in SI within
venous segments interrogated with dynamic HASTE imaging with and
without venous thrombus. A single value of percentage change in SI of
11,000 has been excluded to facilitate the visualization of the results.
Fig. 3. Box-and-whisker plot showing the CAL change in venous
segments interrogated with dynamic HASTE imaging with and without
622 I. Pedrosa et al. / Magnetic Resonance Imaging 27 (2009) 617–624
Fig.6. ROCplot for 117 thresholds of signalchange(13%–1180%by 10 increments): Optimal thresholdof 113% increasein SI (arrow) witha sensitivity of 85%
(95% CI=64%–100%) and a specificity of 76% (95% CI=60%–92%).
Fig. 5. ROC plot from 120 thresholds of CAL change (0–12 mm by 0.1 increments): Optimal threshold of 1.5 mm (arrow) with a sensitivity of 100% and a
specificity of 93% (95% CI=86%–100%).
623 I. Pedrosa et al. / Magnetic Resonance Imaging 27 (2009) 617–624
SI change may require a more vigorous Valsalva Download full-text
maneuver than that necessary to achieve a demonstrable
CAL change. We did not evaluate the patients' ability to
perform a vigorous Valsalva maneuver, and it is possible that
patients with patent veins who did not show any change in
CAL did not exert themselves as much as others who did
show changes in CAL and SI. Correlation with the patients'
ability to perform a strong Valsalva maneuver may provide
insight for further refinements to this technique.
Inherent limitations to this study arise from the retro-
spective, nonblinded nature of the study design, with all
patients known as being referred for possible DVT. With an
interval between the exam date and the image review greater
than 3 months in a busy MR practice, we sought to reduce
the chances of recall bias, although we recognize that it may
have not been eliminated. Also, the number of patients with
thrombosis was small and the number and location of veins
evaluated were heterogeneous. This study represents an
initial observation of MR findings using a dynamic HASTE
imaging approach in patients with suspected venous
thrombosis. A blinded, prospective assessment of this
technique may be valuable in further analyzing the value
of the dynamic HASTE technique for DVT.
Our preliminary experience suggests that the HASTE
sequence paired with the Valsalva maneuver can provide
information about the patency of veins in a short period. The
were excellent when the change in CAL was used. Further
study is required to determine if it has sufficient sensitivity,
specificity and clinical utility to replace other techniques.
 Kakkar VV, Howe CT, Flanc C, Clarke MB. Natural history of
postoperative deep-vein thrombosis. Lancet 1969;2:230–2.
 Fraser JD, Anderson DR. Deep venous thrombosis: recent advances
and optimal investigation with US. Radiology 1999;211:9–24.
 Kalebo P, Anthmyr BA, Eriksson BI, Zachrisson BE. Phlebographic
findings in venous thrombosis following total hip replacement. Acta
 Salzman EW. The epidemiology, pathogenesis, and natural history of
venous thrombosis. In: Colman RWHJ, Marder VJ, Salzman EW,
editors. Hemostasis and thrombosis: basic principles and clinical
practice. Philadelphia: Lippincott; 1995. p. 1275–97.
 Mussurakis S, Papaioannou S, Voros D, Vrakatselis T. Compression
ultrasonography as a reliable imaging monitor in deep venous
thrombosis. Surg Gynecol Obstet 1990;171:233–9.
 Hightower DR, Gooding GA. Sonographic evaluation of the normal
response of subclavian veins to respiratory maneuvers. Invest Radiol
 Effeney DJ, Friedman MB, Gooding GA. Iliofemoral venous
thrombosis: real-time ultrasound diagnosis, normal criteria, and
clinical application. Radiology 1984;150:787–92.
 Raghavendra BN, Rosen RJ, Lam S, Riles T, Horii SC. Deep venous
thrombosis: detection by high-resolution real-time ultrasonography.
 Cronan JJ. Venous thromboembolic disease: the role of US. Radiology
 AndrewsJr EJ,FleischerAC. Sonographyfor deep venousthrombosis:
current and future applications. Ultrasound 2005;21:213–25.
 Fraser DG, Moody AR, Davidson IR, Martel AL, Morgan PS. Deep
venous thrombosis: diagnosis by using venous enhanced subtracted
peak arterial MR venography versus conventional venography.
 Thornton MJ, Ryan R, Varghese JC, Farrell MA, Lucey B, Lee MJ.
A three-dimensional gadolinium-enhanced MR venography techni-
que for imaging central veins. AJR Am J Roentgenol 1999;73:
 Kroencke TJ, Taupitz M, Arnold R, Fritsche L, Hamm B. Three-
dimensional gadolinium-enhanced magnetic resonance venography in
suspected thrombo-occlusive disease of the central chest veins. Chest
 Shinde TS, Lee VS, Rofsky NM, Krinsky GA, Weinreb JC. Three-
dimensional gadolinium-enhanced MR venographic evaluation of
patency of central veins in the thorax: initial experience. Radiology
 Semelka RC, Kelekis NL, Thomasson D, Brown MA, Laub GA.
HASTE MR imaging: description of technique and preliminary results
in the abdomen. J Magn Reson Imaging 1996;6:698–9.
 Regan F. Clinical applications of half-Fourier (HASTE) MR sequences
in abdominal imaging. Magn Reson Imaging Clin N Am 1999;7:
 Gufler H, Laubenberger J, DeGregorio G, Dohnicht S, Langer M.
Pelvic floor descent: dynamic MR imaging using a half-Fourier RARE
sequence. J Magn Reson Imaging 1999;9:378–83.
 Lombardo DM, Baker ME, Spritzer CE, Blinder R, Meyers W,
Herfkens RJ. Hepatic hemangiomas vs metastases: MR differentiation
at 1.5 T. AJR Am J Roentgenol 1990;155:55–9.
 Yu JS, Kim KW, Kim YH, Jeong EK, Chien D. Comparison of
multishotturbospin echoandHASTEsequences for T2-weightedMRI
of liver lesions. J Magn Reson Imaging 1998;8:1079–84.
 Earls JP, Rofsky NM, DeCorato DR, Krinsky GA, Weinreb JC. Breath-
hold single-dose gadolinium-enhanced three-dimensional MR aorto-
graphy: usefulness of a timing examination and MR power injector.
 Lebowitz JA, Rofsky NM, Krinsky GA, Weinreb JC. Gadolinium-
enhanced body MR venography with subtraction technique. AJR Am J
 Laird N, Ware J. Random-effects models for longitudinal data.
 Zhou X, Obuchowski N, McClish D. Statistical methods in diagnostic
medicine. Pharm Stat 2002;2:304–5.
 Finn JP, Zisk JH, Edelman RR, et al. Central venous occlusion: MR
angiography. Radiology 1993;187:245–51.
 Hartnell GG, Hughes LA, Finn JP, Longmaid III HE. Magnetic
resonance angiography of the central chest veins. A new gold
standard? Chest 1995;107:1053–7.
 Ayanzen RH, Bird CR, Keller PJ, McCully FJ, Theobald MR,
Heiserman JE. Cerebral MR venography: normal anatomy and
potential diagnostic pitfalls. AJNR Am J Neuroradiol 2000;21:
 Laub GA. Time-of-flight method of MR angiography. Magn Reson
Imaging Clin N Am 1995;3:391–8.
 Pedrosa I, Morrin M, Oleaga L, Baptista J, Rofsky NM. Is true FISP
imaging reliable in the evaluation of venous thrombosis? J Am J
 Olsen WL, Kucharczyk W, Keyes WD, Norman D, Newton TH.
Magnetic resonance characterization of non-flowing intravascular
blood. Acta Radiol Suppl 1986;369:63–6.
 Barth M, Moser E. Proton NMR relaxation times of human blood
samples at 1.5 T and implications for functional MRI. Cell Mol Biol
 Appelman PT, De Jong TE, Lampmann LE. Deep venous thrombosis
of the leg: US findings. Radiology 1987;163:743–6.
 Gao J, Kazam E, Rubenstein W. The use of deep inspiration preceding
the Valsalva maneuver and rapid expiration in color Doppler imaging
of the lower extremity veins. Clin Imaging 1993;17:266–8.
624I. Pedrosa et al. / Magnetic Resonance Imaging 27 (2009) 617–624