Downloaded from www.ajronline.org by 113.163.2.92 on 10/13/15 from IP address 113.163.2.92. Copyright ARRS. For personal use only; all rights reserved
175
Diagnosis of Femoropopliteal Venous Thrombosis with MR Imaging: A Comparison of Four MR Pulse Sequences
Saara Totterrnan1 Charles W. Francis2 Thomas H. Foster’s Benjamin Brenner Victor J. Marder Robert G. Bryant4
In a prospective
study,
MR
images
were
evaluated
in seven
patients
with
femoro-
popliteal venous thrombosis with symptoms of less than 5 days duration. Ti-weighted (600/25 [TR/TEJ), intermediate (2000/30), and T2-weighted (2000/100) spin-echo series and a gradient-recalled acquisition in the steady state (GRASS) series were compared. Using venography as the standard for diagnosis, we found GRASS to be the most sensitive of the MR techniques, showing thrombi in all patients. It provided good contrast between
the
low-intensity
thrombus
and
high-intensity
flowing
blood
and
also
between
thrombus and intermediateor high-intensity perivascular tissues. The Ti-weighted series was the least sensitive technique. All thrombi showed heterogeneity in the transaxial image with differences in signal between the peripheral and central regions. A higher intensity signal in the center than in the periphery at some level of the thrombus was found in six of seven T2-weighted or GRASS images. Heterogeneity in the signal intensity was more frequent in distal portions of thrombi, whereas the most proximal extent was homogeneous in appearance in six of seven cases. The heterogeneous appearance may be related to the greater age of the distal thrombus, because deep venous thrombi are known to begin in the calf and extend proximally over time. We conclude, on the basis of our experience with a small number of patients, that the GRASS MR technique is more sensitive for detecting acute deep venous thrombosis than Ti-weighted, intermediate, and T2-weighted MR images. AJR
154:175-178,
January
1990
Thrombi sequences acquisition Received
June 9, 1989;
accepted
after revision
July 28, 1989. This work was supported in part by grant HL3061 6 from the National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda,
MD. C. W. Francis is the recipient of an Established Investigator Award from the American Heart Association with funds provided by the New Vork State Affiliate. Department of Radiology, P.O. versity of Rochester Medical Center, I
Ave., Rochester,
NV 14642.
quests to S. Totterman. 2 Department of Medicine,
Box 694, Uni601 Elmwood Address reprint reUniversity
chester Medical Center, Rochester, 3
Department
of Physics
sity of Rochester, 4
Department
chester, Rochester,
Univer-
NV 14627.
of Biophysics,
NV 14642.
0361 -803X/90/1 541-01 75 © American Roentgen Ray Society
University
ent
levels.
chosen provide needed
Only
patients
with
thrombosis
involving
femoral
or popliteal
veins
were
for imaging because our pilot studies showed that the body coil did not sufficient signal-to-noise ratio for high-resolution, small-field-of-view images to assess the internal structure of venous thrombi in the calf.
Ro-
NV 14642.
and Astronomy,
Rochester,
of
in abdominal and pelvic veins have been identified by using spin-echo with Ti - or T2-weighted MR images [1 -4], and a gradient-recalled in the steady state (GRASS) sequence has been used to identify thrombi in iliac and femoral veins [5, 6]. The accuracy of MR imaging in diagnosing deep venous thrombosis and its ability to distinguish age-related properties of thrombi may be related to the choice of image acquisition sequence and to the particular acquisition parameters selected. Therefore, we have compared four different acquisition sequences in patients with acute thrombosis of the femoropopliteal veins proved with venography, evaluating the cross-sectional appearance at differ-
of Ro-
Subjects
and Methods
MR imaging was performed on seven patients with symptoms of less than S days duration consistent with femoropopliteal deep venous thrombosis in whom venography showed acute thrombus in the calf, the popliteal vein, the superficial femoral vein, and/or the common femoral vein. The patients included four men and three women, from 26 to 71 years old. Informed consent was obtained from all patients.
TOTTERMAN
176
ET
AL.
AJR:i54,
January 1990
Fig. 1.-69-year-old woman with acute yenous thrombosis in calf, popliteal vein, and superficial femoral vein. MR images show typical appearance of thrombi at two levels in thigh with four
different
pulse
sequences.
Axial
images
were obtained at proximal tip of thrombus (A-D) and 10 cm distally
Downloaded from www.ajronline.org by 113.163.2.92 on 10/13/15 from IP address 113.163.2.92. Copyright ARRS. For personal use only; all rights reserved
A-D,
(E-H).
Gradient-recalled
acquisition
in
the
steady state (GRASS) image (A) shows lowintensity thrombus (arrow) easily distinguishable from high-intensityflowing blood. 600/25 (B) and 2000/30 (C) images show intermediate-intensity
thrombus,
which is difficult to distinguish from image (D) shows areas
flowing blood. 2000/100 of high and low signal
intensity,
and thrombus
cannot be identified. appears
as a low-intensity
thrombus (arrow) area but has inter-
mediate
intensity
(F) and 2000/30
E-H,
In GRASS
image in 600/20
(E),
(G)
images.
High-intensity centers are shown in 2000/30 (G) and 2000/100 (H) images. Edematees penvascular tissue appears as a high-signal-intensity area (arrows) in H.
The
were conducted with a i .5-T General Electric 24 hr of venography. Axial images were first obtained with a GRASS sequence with first-order flow compensation, by using 1 0-mm slices and 1 0-mm interslice intervals from the level of mid pelvis to the knee. Image data were acquired with a TRITE of 33/i7, flip angles of 50#{176}-70#{176}, two excitations, and a i28 x 256 Signa
MR studies
irnager
matrix.
within
Ti -weighted
images
were
obtained
at 600/20-30,
intermedi-
ate images at 2000/30, and T2-weighted images at 2000/i 00 with the same imaging parameters. The body coil was used as transmitter for all patients and as receiver for five. For two patients in whom venography
showed
cial femoral coil
was
used
vein,
thrombosis
extending
a homemade,
as a receiver.
only
single-loop, The
field
of view
to the
distal
saddle-shaped was
adjusted
superfi-
surface on
the
of the size of the patient; it varied from 20 to 34 cm for images obtained with the body coil and was 20 cm for surface-coil images. The images were evaluated for contrast between thrombus and flowing blood or perivascular tissues and also for relative signal intensity in the center compared with the periphery of the thrombus at different levels. The effect of saturation pulses and of first-order flow compensation on spin-echo images of proximal veins in the thigh was tested in two volunteers. basis
Results The femoropopliteal veins could be localized without culty with all four acquisition sequences in areas where
diffithere
MR
January 1990
AJR:154,
OF
FEMOROPOPLITEAL
VENOUS
THROMBOSIS
Ti -weighted
images were less useful in identifying thrombi blood appeared most often as an area of intermediate signal intensity or showed varying intensity rims or a flow-void, depending on the slice location (Fig. 2B). Most thrombi at all levels appeared as intermediate-intensity homogeneous areas, and little contrast was seen between thrombus and flowing blood or perivascular tissues. Heterogeneity in the thrombus was seen on Ti -weighted images in only three patients as opposed to the GRASS images, which showed more image variability. The appearance of thrombi obtained at 2000/30 was similar to that obtained with the Ti -weighted sequence (Fig. i C). Flowing blood appeared as a flow-void, an area of intermediate signal intensity (Fig. 2C), or had different intensity rims, depending on slice location. The appearance of the proximal portion of the thrombus did not differ from the flowing blood in the contralateral vein; both showed intermediate signal intensity (six patients) or a high-intensity rim (one patient). More heterogeneity in the image appeared in the distal portion of the thrombus when the intermediate pulse sequence was used compared with the Ti -weighted images, with five thrombi showing higher intensity centers compared with the periphery of the clot at some level (Fig. i G). In T2-weighted images, the most proximal portion of the thrombus had intermediate intensity but could not be distinguished from signal originating from flowing blood (Fig. i D). The most proximal portion of the thrombus in one patient was unusual in showing a low-intensity rim with an intermediate-intensity center, similar to its appearance with the GRASS sequence. However, this pattern was seen distally with T2-weighted images at one or more locations in all of the other patients except for one in whom the thrombus was homogeneous throughout. The GRASS sequence was the most sensitive in showing heterogeneity in the cross-sectional appearance of thrombi, with six of seven having a higher intensity center in one or more sections distal to the tip. In four of seven patients, sections 2-3 cm distal from the thrombus tip showed heterogeneity. Six of seven patients showed this pattern 5-6 cm distal from the thrombus tip. The T2-weighted image showed heterogeneity in the thrombus in six of seven patients, but this occurred in fewer levels than seen with the GRASS sequence. In three patients, all sequences showed a higher intensity center in at least one section. A higher intensity periphery was seen less commonly; it was found in only two patients with the intermediate sequence and in one patient with the T2-weighted sequence. Studies in volunteers showed that the signal from flowing blood in proximal veins in the thigh in Ti-weighted, intermediate, and T2-weighted spinecho images varied from low to high intensity whether or not saturation pulses or first-order flow compensation was used. Because the GRASS images were the most sensitive, identifying thrombi in all patients, they were compared with the venographic findings. Flowing blood could be identified in all cases in which a thrombus filled 60% or less of the crosssectional area of the vein, and the most proximal portion of the thrombus was easier to recognize in transaxial GRASS images than in the venogram. One thrombus nearly filled the
Downloaded from www.ajronline.org by 113.163.2.92 on 10/13/15 from IP address 113.163.2.92. Copyright ARRS. For personal use only; all rights reserved
(Fig.
B
C
D Fig. 2.-26-year-old
vein,
and
superficial
man with acute venous thrombosis femoral
vein.
MR images
in calf, popliteal
obtained
with
four
pulse
sequences show heterogeneity in thrombus. A-D, Axial MR images were obtained 10 cm from proximal tip of thrombus. In GRASS image (A), thrombus (arrow) has a low-intensity peripheral rim with higher
signal
intermediate-intensity blood.
In both
2000/30
higher signal intensity
was
sufficient
groin,
upper
in center.
In 600/25
area and cannot (C)
and
2000/100
image
(B),
it appears
be distinguished (D)
images,
as an
from flowing
periphery
has
a
than center.
adjacent adipose tissue (e.g., in the pelvis, and popliteal fossa). Identification was more difficult in lean patients or in the lower thigh and calf, where there was less perivascular fat. The appearance of thrombus varied depending on the pulse sequence used and varied in distal and proximal portions. In GRASS images, the tip of the thrombus had a homogeneous low intensity (Fig. 1 A) in five patients, intermediate intensity in one patient, or had a low intensity “rim” with an intermediate center in one patient (Fig. 2A). There was good contrast between thrombus and the high-intensity signal from flowing blood in all cases. Distal from the tip, the thrombus in only one patient had a homogeneous low intensity (Fig. 1 E); all others showed a low-intensity rim with intermediate center (Fig. 2A). This appearance was distinct from flowing blood or surrounding perivascular tissues, permitting definite identification of the clot. thigh,
177
1 B). Flowing
178
TOTTERMAN
and was surrounded by a narrow rim of contrast on venography; this peripheral rim of flowing blood was not seen in GRASS images. The venogram and GRASS image were in agreement regarding the presence of thrombus in all cases in the external iliac, common femoral, superficial femoral, and popliteal veins.
Downloaded from www.ajronline.org by 113.163.2.92 on 10/13/15 from IP address 113.163.2.92. Copyright ARRS. For personal use only; all rights reserved
vein
Discussion Analysis of the MR images showed that the GRASS sequence was better than Ti -weighted, intermediate, and T2weighted images in identifying thrombus in femoral and popliteal veins. With this sequence, image slices are acquired sequentially rather than in an interleaved fashion. Thus, protons in flowing blood in any particular slice are continuously replaced with fully magnetized spins, while the surrounding stationary tissue is saturated [7]. The time-of-flight effect of flow is responsible for the efficiency of this approach and makes the GRASS sequence more effective than Ti - or T2weighted sequences. Also, the GRASS sequence provides good contrast between flowing blood and thrombus, as shown previously [5]. Although excellent agreement was seen between GRASS images and venograms in identification of thrombus in the popliteal vein and more proximal veins in the leg, the signal-to-noise ratio available with the body coil used did not support the small-field-of-view images that are needed for more detailed evaluation of the structure of venous thrombi in the calf. Although the use of specialized surface coils could solve this problem, multiple imaging coils and sequences would be required to evaluate the veins in the leg. In spin-echo images, many factors contribute to the signal intensity of flow, including the distance of the image slice from the entry point of the flowing blood; the TRITE combination; the number of echoes; and the direction, velocity, and character of the flow [8-i 0]. Our experience shows that spinecho images with Ti -weighting, T2-weighting, or intermediate weighting were diagnostically inferior to the GRASS Sequence, providing little contrast between thrombus and surrounding tissue or flowing blood, and failing to show thrombus at many sites. On the basis of results of studies performed on normal subjects, we did not use presaturation pulses or first-order flow compensation in the spin-echo examinations. The MR appearance of flowing blood in veins in the groin showed considerable variation whether or not these methods were used. The reason for the lack of a consistent flow-void with presaturation pulses and for the lack of high signal from flowing blood with flow compensation in these veins is not clear, but it may be related to the low velocity of venous flow. The spin-echo sequences, particularly the T2-weighted images, were adequate for evaluating inflammatory changes in tissues next to thrombosed veins. Variations in the MR image of thrombi may reflect agerelated changes in the composition of the thrombi that are of
ET AL.
AJR:154, January 1990
potential diagnostic importance. Prior studies that used spinecho sequences have shown differences in the MR image of acute compared with older venous thrombi [i i -i 3]. Our studies also show variations in MR appearance of acute venous thrombi, showing heterogeneity in more distal portions. This heterogeneity may be age-related, because most thrombi are thought to begin in the calf and then extend over a variable period of time to involve more proximal veins [i 4]. Therefore, the most proximal portion of the thrombus may be regarded as the most recent and distal parts as older clot. This is consistent with our findings, in which the tip of the thrombus representing recent clot was homogeneous, whereas distal, older portions were heterogeneous in appearance with a periphery of lower intensity and a center of higher image intensity.
ACKNOWLEDGMENTS The authors
thank
Connie
assistance
with this study
preparation
of the manuscript.
White
and Christina Keough for their Weed for her help in the
and Carol
REFERENCES 1 . Kneeland JB, Auh VH, zirinsky
K, Rubenstein W, Kazam E. MA, CT, and ultrasonographic demonstration of splenic vein thrombosis. J Comput Assist Tomogr 1984;8: 1 199-1 200 2. Hricak H, Amparo E, Fisher MA, Crooks LE, Higgins CB. Abdominal venous system: assessment using MR. Radiology 1985;156:415-422 3. Bernardino ME, Steinberg HV, Pearson TC, Gedgaudas-McClees AK, Torres WE, angiography
Henderson JM. for determination
4. Martin B, Mulopulos
Shunts for portal hypertension: MR of patency. Radiology 1986;158:57-61
GP, Bryan PJ. MRI of puerperal ovarian-vein
bosis. AJR 1986;147:291-292 5. Spritzer CE, Sussman 5K, Blinder venous thrombosis evaluation with
and
throm-
RA, Saeed M, Herfkens RJ. Deep limited-flip-angle, gradient-refocused Radiology 1988;166:371-375
MR imaging: preliminary experience. 6. Francis CW, Foster TH, Totterman 5, Brenner B, Marder VJ, Bryant AG. Monitoring of therapy for deep vein thrombosis using imaging in two cases. Acta Radiologica (in press)
magnetic
resonance
7. Winkler ML, Ortendahl DA, Mills TC, et al. Characteristics of partial flip angle and gradient reversal MR imaging. Radiology 1988;166: 17-26 8. Bradley WG Jr, Waluch V, Lai K-S, Fernandez EJ, Spalter C. The appearance of rapidly flowing blood on magnetic resonance images. AJR 1984;143:1167-1174
9. Kucharczyk W, Kelly WM, Davis DO, Norman D, Newton TH. Intracranial lesions: flow-related enhancement on MR images using time-of-flight offects. Radiology 1986;161 :767-772 10. Bradley WF Jr. Flow phenomena in MR imaging. AJR 1988;150:983-994 1 1 . Macchi Bilaniuk Comput
PJ, Grossman RI, Gomori JM, Goldberg LT. High field MR imaging of cerebral Assist Tomogr 1986;1 0:10-15
12. Atlas SW, Grossman man RA. Partially
HI, Zimmerman venous thrombosis.
RA, J
RI, Goldberg HI, Hackney DB, Bilaniuk LT, Zimmer-
thrombosed
giant
intracranial
aneurysms:
correlation
of
MR and pathologic findings. Radiology 1987;1 62:111-114 13. Zirinsky K, Markisz JA, Rubinstein WA, et al. MR imaging of portal venous thrombosis: correlation with CT and sonography. AJR 1988;i50:283-288 14. Kakkar VV, Flanc C, Howe CT, Clarke MB. Natural history of postoperative deep-vein thrombosis. Lancet 1969;2:230-232
175
Diagnosis of Femoropopliteal Venous Thrombosis with MR Imaging: A Comparison of Four MR Pulse Sequences
Saara Totterrnan1 Charles W. Francis2 Thomas H. Foster’s Benjamin Brenner Victor J. Marder Robert G. Bryant4
In a prospective
study,
MR
images
were
evaluated
in seven
patients
with
femoro-
popliteal venous thrombosis with symptoms of less than 5 days duration. Ti-weighted (600/25 [TR/TEJ), intermediate (2000/30), and T2-weighted (2000/100) spin-echo series and a gradient-recalled acquisition in the steady state (GRASS) series were compared. Using venography as the standard for diagnosis, we found GRASS to be the most sensitive of the MR techniques, showing thrombi in all patients. It provided good contrast between
the
low-intensity
thrombus
and
high-intensity
flowing
blood
and
also
between
thrombus and intermediateor high-intensity perivascular tissues. The Ti-weighted series was the least sensitive technique. All thrombi showed heterogeneity in the transaxial image with differences in signal between the peripheral and central regions. A higher intensity signal in the center than in the periphery at some level of the thrombus was found in six of seven T2-weighted or GRASS images. Heterogeneity in the signal intensity was more frequent in distal portions of thrombi, whereas the most proximal extent was homogeneous in appearance in six of seven cases. The heterogeneous appearance may be related to the greater age of the distal thrombus, because deep venous thrombi are known to begin in the calf and extend proximally over time. We conclude, on the basis of our experience with a small number of patients, that the GRASS MR technique is more sensitive for detecting acute deep venous thrombosis than Ti-weighted, intermediate, and T2-weighted MR images. AJR
154:175-178,
January
1990
Thrombi sequences acquisition Received
June 9, 1989;
accepted
after revision
July 28, 1989. This work was supported in part by grant HL3061 6 from the National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda,
MD. C. W. Francis is the recipient of an Established Investigator Award from the American Heart Association with funds provided by the New Vork State Affiliate. Department of Radiology, P.O. versity of Rochester Medical Center, I
Ave., Rochester,
NV 14642.
quests to S. Totterman. 2 Department of Medicine,
Box 694, Uni601 Elmwood Address reprint reUniversity
chester Medical Center, Rochester, 3
Department
of Physics
sity of Rochester, 4
Department
chester, Rochester,
Univer-
NV 14627.
of Biophysics,
NV 14642.
0361 -803X/90/1 541-01 75 © American Roentgen Ray Society
University
ent
levels.
chosen provide needed
Only
patients
with
thrombosis
involving
femoral
or popliteal
veins
were
for imaging because our pilot studies showed that the body coil did not sufficient signal-to-noise ratio for high-resolution, small-field-of-view images to assess the internal structure of venous thrombi in the calf.
Ro-
NV 14642.
and Astronomy,
Rochester,
of
in abdominal and pelvic veins have been identified by using spin-echo with Ti - or T2-weighted MR images [1 -4], and a gradient-recalled in the steady state (GRASS) sequence has been used to identify thrombi in iliac and femoral veins [5, 6]. The accuracy of MR imaging in diagnosing deep venous thrombosis and its ability to distinguish age-related properties of thrombi may be related to the choice of image acquisition sequence and to the particular acquisition parameters selected. Therefore, we have compared four different acquisition sequences in patients with acute thrombosis of the femoropopliteal veins proved with venography, evaluating the cross-sectional appearance at differ-
of Ro-
Subjects
and Methods
MR imaging was performed on seven patients with symptoms of less than S days duration consistent with femoropopliteal deep venous thrombosis in whom venography showed acute thrombus in the calf, the popliteal vein, the superficial femoral vein, and/or the common femoral vein. The patients included four men and three women, from 26 to 71 years old. Informed consent was obtained from all patients.
TOTTERMAN
176
ET
AL.
AJR:i54,
January 1990
Fig. 1.-69-year-old woman with acute yenous thrombosis in calf, popliteal vein, and superficial femoral vein. MR images show typical appearance of thrombi at two levels in thigh with four
different
pulse
sequences.
Axial
images
were obtained at proximal tip of thrombus (A-D) and 10 cm distally
Downloaded from www.ajronline.org by 113.163.2.92 on 10/13/15 from IP address 113.163.2.92. Copyright ARRS. For personal use only; all rights reserved
A-D,
(E-H).
Gradient-recalled
acquisition
in
the
steady state (GRASS) image (A) shows lowintensity thrombus (arrow) easily distinguishable from high-intensityflowing blood. 600/25 (B) and 2000/30 (C) images show intermediate-intensity
thrombus,
which is difficult to distinguish from image (D) shows areas
flowing blood. 2000/100 of high and low signal
intensity,
and thrombus
cannot be identified. appears
as a low-intensity
thrombus (arrow) area but has inter-
mediate
intensity
(F) and 2000/30
E-H,
In GRASS
image in 600/20
(E),
(G)
images.
High-intensity centers are shown in 2000/30 (G) and 2000/100 (H) images. Edematees penvascular tissue appears as a high-signal-intensity area (arrows) in H.
The
were conducted with a i .5-T General Electric 24 hr of venography. Axial images were first obtained with a GRASS sequence with first-order flow compensation, by using 1 0-mm slices and 1 0-mm interslice intervals from the level of mid pelvis to the knee. Image data were acquired with a TRITE of 33/i7, flip angles of 50#{176}-70#{176}, two excitations, and a i28 x 256 Signa
MR studies
irnager
matrix.
within
Ti -weighted
images
were
obtained
at 600/20-30,
intermedi-
ate images at 2000/30, and T2-weighted images at 2000/i 00 with the same imaging parameters. The body coil was used as transmitter for all patients and as receiver for five. For two patients in whom venography
showed
cial femoral coil
was
used
vein,
thrombosis
extending
a homemade,
as a receiver.
only
single-loop, The
field
of view
to the
distal
saddle-shaped was
adjusted
superfi-
surface on
the
of the size of the patient; it varied from 20 to 34 cm for images obtained with the body coil and was 20 cm for surface-coil images. The images were evaluated for contrast between thrombus and flowing blood or perivascular tissues and also for relative signal intensity in the center compared with the periphery of the thrombus at different levels. The effect of saturation pulses and of first-order flow compensation on spin-echo images of proximal veins in the thigh was tested in two volunteers. basis
Results The femoropopliteal veins could be localized without culty with all four acquisition sequences in areas where
diffithere
MR
January 1990
AJR:154,
OF
FEMOROPOPLITEAL
VENOUS
THROMBOSIS
Ti -weighted
images were less useful in identifying thrombi blood appeared most often as an area of intermediate signal intensity or showed varying intensity rims or a flow-void, depending on the slice location (Fig. 2B). Most thrombi at all levels appeared as intermediate-intensity homogeneous areas, and little contrast was seen between thrombus and flowing blood or perivascular tissues. Heterogeneity in the thrombus was seen on Ti -weighted images in only three patients as opposed to the GRASS images, which showed more image variability. The appearance of thrombi obtained at 2000/30 was similar to that obtained with the Ti -weighted sequence (Fig. i C). Flowing blood appeared as a flow-void, an area of intermediate signal intensity (Fig. 2C), or had different intensity rims, depending on slice location. The appearance of the proximal portion of the thrombus did not differ from the flowing blood in the contralateral vein; both showed intermediate signal intensity (six patients) or a high-intensity rim (one patient). More heterogeneity in the image appeared in the distal portion of the thrombus when the intermediate pulse sequence was used compared with the Ti -weighted images, with five thrombi showing higher intensity centers compared with the periphery of the clot at some level (Fig. i G). In T2-weighted images, the most proximal portion of the thrombus had intermediate intensity but could not be distinguished from signal originating from flowing blood (Fig. i D). The most proximal portion of the thrombus in one patient was unusual in showing a low-intensity rim with an intermediate-intensity center, similar to its appearance with the GRASS sequence. However, this pattern was seen distally with T2-weighted images at one or more locations in all of the other patients except for one in whom the thrombus was homogeneous throughout. The GRASS sequence was the most sensitive in showing heterogeneity in the cross-sectional appearance of thrombi, with six of seven having a higher intensity center in one or more sections distal to the tip. In four of seven patients, sections 2-3 cm distal from the thrombus tip showed heterogeneity. Six of seven patients showed this pattern 5-6 cm distal from the thrombus tip. The T2-weighted image showed heterogeneity in the thrombus in six of seven patients, but this occurred in fewer levels than seen with the GRASS sequence. In three patients, all sequences showed a higher intensity center in at least one section. A higher intensity periphery was seen less commonly; it was found in only two patients with the intermediate sequence and in one patient with the T2-weighted sequence. Studies in volunteers showed that the signal from flowing blood in proximal veins in the thigh in Ti-weighted, intermediate, and T2-weighted spinecho images varied from low to high intensity whether or not saturation pulses or first-order flow compensation was used. Because the GRASS images were the most sensitive, identifying thrombi in all patients, they were compared with the venographic findings. Flowing blood could be identified in all cases in which a thrombus filled 60% or less of the crosssectional area of the vein, and the most proximal portion of the thrombus was easier to recognize in transaxial GRASS images than in the venogram. One thrombus nearly filled the
Downloaded from www.ajronline.org by 113.163.2.92 on 10/13/15 from IP address 113.163.2.92. Copyright ARRS. For personal use only; all rights reserved
(Fig.
B
C
D Fig. 2.-26-year-old
vein,
and
superficial
man with acute venous thrombosis femoral
vein.
MR images
in calf, popliteal
obtained
with
four
pulse
sequences show heterogeneity in thrombus. A-D, Axial MR images were obtained 10 cm from proximal tip of thrombus. In GRASS image (A), thrombus (arrow) has a low-intensity peripheral rim with higher
signal
intermediate-intensity blood.
In both
2000/30
higher signal intensity
was
sufficient
groin,
upper
in center.
In 600/25
area and cannot (C)
and
2000/100
image
(B),
it appears
be distinguished (D)
images,
as an
from flowing
periphery
has
a
than center.
adjacent adipose tissue (e.g., in the pelvis, and popliteal fossa). Identification was more difficult in lean patients or in the lower thigh and calf, where there was less perivascular fat. The appearance of thrombus varied depending on the pulse sequence used and varied in distal and proximal portions. In GRASS images, the tip of the thrombus had a homogeneous low intensity (Fig. 1 A) in five patients, intermediate intensity in one patient, or had a low intensity “rim” with an intermediate center in one patient (Fig. 2A). There was good contrast between thrombus and the high-intensity signal from flowing blood in all cases. Distal from the tip, the thrombus in only one patient had a homogeneous low intensity (Fig. 1 E); all others showed a low-intensity rim with intermediate center (Fig. 2A). This appearance was distinct from flowing blood or surrounding perivascular tissues, permitting definite identification of the clot. thigh,
177
1 B). Flowing
178
TOTTERMAN
and was surrounded by a narrow rim of contrast on venography; this peripheral rim of flowing blood was not seen in GRASS images. The venogram and GRASS image were in agreement regarding the presence of thrombus in all cases in the external iliac, common femoral, superficial femoral, and popliteal veins.
Downloaded from www.ajronline.org by 113.163.2.92 on 10/13/15 from IP address 113.163.2.92. Copyright ARRS. For personal use only; all rights reserved
vein
Discussion Analysis of the MR images showed that the GRASS sequence was better than Ti -weighted, intermediate, and T2weighted images in identifying thrombus in femoral and popliteal veins. With this sequence, image slices are acquired sequentially rather than in an interleaved fashion. Thus, protons in flowing blood in any particular slice are continuously replaced with fully magnetized spins, while the surrounding stationary tissue is saturated [7]. The time-of-flight effect of flow is responsible for the efficiency of this approach and makes the GRASS sequence more effective than Ti - or T2weighted sequences. Also, the GRASS sequence provides good contrast between flowing blood and thrombus, as shown previously [5]. Although excellent agreement was seen between GRASS images and venograms in identification of thrombus in the popliteal vein and more proximal veins in the leg, the signal-to-noise ratio available with the body coil used did not support the small-field-of-view images that are needed for more detailed evaluation of the structure of venous thrombi in the calf. Although the use of specialized surface coils could solve this problem, multiple imaging coils and sequences would be required to evaluate the veins in the leg. In spin-echo images, many factors contribute to the signal intensity of flow, including the distance of the image slice from the entry point of the flowing blood; the TRITE combination; the number of echoes; and the direction, velocity, and character of the flow [8-i 0]. Our experience shows that spinecho images with Ti -weighting, T2-weighting, or intermediate weighting were diagnostically inferior to the GRASS Sequence, providing little contrast between thrombus and surrounding tissue or flowing blood, and failing to show thrombus at many sites. On the basis of results of studies performed on normal subjects, we did not use presaturation pulses or first-order flow compensation in the spin-echo examinations. The MR appearance of flowing blood in veins in the groin showed considerable variation whether or not these methods were used. The reason for the lack of a consistent flow-void with presaturation pulses and for the lack of high signal from flowing blood with flow compensation in these veins is not clear, but it may be related to the low velocity of venous flow. The spin-echo sequences, particularly the T2-weighted images, were adequate for evaluating inflammatory changes in tissues next to thrombosed veins. Variations in the MR image of thrombi may reflect agerelated changes in the composition of the thrombi that are of
ET AL.
AJR:154, January 1990
potential diagnostic importance. Prior studies that used spinecho sequences have shown differences in the MR image of acute compared with older venous thrombi [i i -i 3]. Our studies also show variations in MR appearance of acute venous thrombi, showing heterogeneity in more distal portions. This heterogeneity may be age-related, because most thrombi are thought to begin in the calf and then extend over a variable period of time to involve more proximal veins [i 4]. Therefore, the most proximal portion of the thrombus may be regarded as the most recent and distal parts as older clot. This is consistent with our findings, in which the tip of the thrombus representing recent clot was homogeneous, whereas distal, older portions were heterogeneous in appearance with a periphery of lower intensity and a center of higher image intensity.
ACKNOWLEDGMENTS The authors
thank
Connie
assistance
with this study
preparation
of the manuscript.
White
and Christina Keough for their Weed for her help in the
and Carol
REFERENCES 1 . Kneeland JB, Auh VH, zirinsky
K, Rubenstein W, Kazam E. MA, CT, and ultrasonographic demonstration of splenic vein thrombosis. J Comput Assist Tomogr 1984;8: 1 199-1 200 2. Hricak H, Amparo E, Fisher MA, Crooks LE, Higgins CB. Abdominal venous system: assessment using MR. Radiology 1985;156:415-422 3. Bernardino ME, Steinberg HV, Pearson TC, Gedgaudas-McClees AK, Torres WE, angiography
Henderson JM. for determination
4. Martin B, Mulopulos
Shunts for portal hypertension: MR of patency. Radiology 1986;158:57-61
GP, Bryan PJ. MRI of puerperal ovarian-vein
bosis. AJR 1986;147:291-292 5. Spritzer CE, Sussman 5K, Blinder venous thrombosis evaluation with
and
throm-
RA, Saeed M, Herfkens RJ. Deep limited-flip-angle, gradient-refocused Radiology 1988;166:371-375
MR imaging: preliminary experience. 6. Francis CW, Foster TH, Totterman 5, Brenner B, Marder VJ, Bryant AG. Monitoring of therapy for deep vein thrombosis using imaging in two cases. Acta Radiologica (in press)
magnetic
resonance
7. Winkler ML, Ortendahl DA, Mills TC, et al. Characteristics of partial flip angle and gradient reversal MR imaging. Radiology 1988;166: 17-26 8. Bradley WG Jr, Waluch V, Lai K-S, Fernandez EJ, Spalter C. The appearance of rapidly flowing blood on magnetic resonance images. AJR 1984;143:1167-1174
9. Kucharczyk W, Kelly WM, Davis DO, Norman D, Newton TH. Intracranial lesions: flow-related enhancement on MR images using time-of-flight offects. Radiology 1986;161 :767-772 10. Bradley WF Jr. Flow phenomena in MR imaging. AJR 1988;150:983-994 1 1 . Macchi Bilaniuk Comput
PJ, Grossman RI, Gomori JM, Goldberg LT. High field MR imaging of cerebral Assist Tomogr 1986;1 0:10-15
12. Atlas SW, Grossman man RA. Partially
HI, Zimmerman venous thrombosis.
RA, J
RI, Goldberg HI, Hackney DB, Bilaniuk LT, Zimmer-
thrombosed
giant
intracranial
aneurysms:
correlation
of
MR and pathologic findings. Radiology 1987;1 62:111-114 13. Zirinsky K, Markisz JA, Rubinstein WA, et al. MR imaging of portal venous thrombosis: correlation with CT and sonography. AJR 1988;i50:283-288 14. Kakkar VV, Flanc C, Howe CT, Clarke MB. Natural history of postoperative deep-vein thrombosis. Lancet 1969;2:230-232
