Pearls and Pitfalls in the Magnetic Resonance Diagnosis of Dural Sinus Thrombosis: A Comprehensive Guide for the Trainee Radiologist
Dural sinus thrombosis (DST) is a neurological emergency, accounting for 1–2% of all young adult strokes. Its clinical presentation varies with location and subsequent evolution of the thrombus. Common presenting complaints include headache, motor deficits, and convulsions. These symptoms are non-specific and overlap with various other neurological diseases, making the clinical diagnosis of DST challenging and often delaying confirmatory investigations.[2,3] With prompt treatment, DST can be reversed, but if untreated, it may rapidly progress to brain parenchymal infarction, parenchymal hemorrhages, altered mentation, coma, and death.
Radiology is the investigation of choice for the detection of DST. Non-enhanced CT (NECT) is often the initial investigation, mainly due to its widespread availability and quick scan time. The classic hyperdense appearance of DST on NECT is seen only in a third of all patients, making the sensitivity of NECT poor.[5,6] CT Venogram is often used as a complimentary tool to NECT with a reported sensitivity of 95% in detecting DST. It requires iodinated contrast media to delineate a non-enhancing thrombus within the opacified venous system.[1,4] CT contrast media pose an increased risk of allergic reactions and nephrotoxicity. Furthermore, CT in view of its ionizing radiation is best avoided in pregnancy, an important group of patients vulnerable to DST.
Over the years, Magnetic Resonance Imaging (MRI) has been used more frequently for the diagnosis of DST and its complications.[9,10] MRI offers multiple advantages which include its non-ionizing property and excellent contrast resolution that help to identify and age DST accurately. Patel et al. reported a sensitivity and specificity of 79.2% and 89.9%, respectively, of routine MRI sequences such as spin- and gradient-echo, diffusion-weighted imaging, and inversion recovery in the diagnosis of DST. In addition to these MR sequences, multiple MR venography (MRV) techniques are routinely used to evaluate and quantify flow within the dural sinuses. MR venograms can be performed with or without gadolinium administration. MRV has an overall sensitivity and negative predictive value of 93.54% and 90.9%, respectively.
Albeit its advantages over CT, MRI suffers a few limitations. Physiological changes in blood flow, temporal evolution of DST, and incorrect selection/application of MR techniques are common sources of erroneous assumption of sinovenous occlusion or patency. These can lead to an inaccurate diagnosis or a missed pathology.
This educational review provides a simplified understanding of the basic principles of commonly used MRI techniques in the diagnosis of DST. We have highlighted common pitfalls encountered with the use of these techniques along with troubleshooting methods, using case-based imaging examples.
BASIC PRINCIPLES IN THE MR EVALUATION OF DURAL SINUSES
Spin echo imaging (SEI)
Conventional SEI utilizes two pulses for signal generation; first, a 90° radiofrequency (RF) pulse to tip the magnetization vector, followed by a 180° rephasing pulse. Mobile spins fail to experience the rephasing pulse and hence do not exhibit a signal. Therefore, patent dural sinuses demonstrate “flow voids” or profound hypointensity [Figure 1].[13,14]
Bland DST primarily comprises red blood cells. As a result, it evolves similar to a parenchymal hematoma. Replacement of a signal void (“loss of flow void”) within a dural sinus on SEI must raise suspicion of DST and prompt further evaluation of the dural sinus.
Time-of-flight (TOF) MRV
TOF imaging relies on fundamental principles of “flow-related enhancement.” The sequence is a gradient echo-based sequence that achieves saturation and hence signals suppression of stationary protons by virtue of short interval RF pulses. By selecting an orientation perpendicular to venous flow, bright blood images are obtained on account of freshly entering unsaturated protons imparting a signal [Figure 2].
Additional saturation bands are applied selectively to nullify signals from adjacent arteries.
Phase-contrast (PC) MRV
Flowing spins change phase (“phase shift”) as they move along a magnetic gradient. PC techniques use this phenomenon to generate images. It uses additional bipolar gradients to initially induce a phase shift and subsequently to restore it. The net phase difference of stationary protons is zero; however, due to spatial changes, the difference is non-zero for mobile spins [Figure 3].[16,17] Voxels in PC imaging therefore represent phase shift data and not signal amplitude, with stationary spins not contributing to the image. Using a flow encoding gradient, also known as velocity encoding gradient (VENC), a threshold can be determined to display arterial or venous flow. Higher VENC visualizes arteries and lower VENC visualizes venous flow and cerebrospinal fluid flow (CSF) circulation.[13,19]
Contrast-enhanced (CE) MRV
Gadolinium chelates are paramagnetic and shorten T1 relaxation time. Following injection, they increase intravascular signal against the background tissues creating a “lumenogram.” CE-MRV is typically a 3D-T1 gradient echo-based sequence with short TR and TE to achieve heavily T1W images.[13,16] Techniques such as test bolus and bolus tracking are used to ensure that images are obtained in the venous phase. Despite the need for contrast media, CE-MRV provides advantages such as fast data acquisition. It also uses thin slices and high matrix which together provide high spatial resolution (independent of flow dynamics) with multiplanar reconstruction capability.[13,20,21] CE-MRV has a reported sensitivity and specificity of 83% and 100%, respectively, in the detection of DST. A comparison of the different MRV techniques and the MRV parameters used in our department have been summarized in Tables 1 and 2 respectively.
|Principle||Flow enhancement||Phase shift||T1 shortening|
|Sequence||Gradient echo-based||Gradient echo-based||Gradient echo-based|
|Acquisition time||5–8 min||>15 min*||Approx. 4 min|
|Background||Partly suppressed||Suppressed||Enhancing lesions seen|
|Factors for accuracy||Saturation of exogenous signal (arteries)||Operator for selection of appropriate VENC||Scan timing for maximum venous enhancement|
|Flow velocity||Can detect slow flow||Depends on the VENC||Independent|
|Flow direction||Not detected||Detected||Not detected|
|Acquisition type||2D (axial and coronal)||3D (sagittal left to right)||3D (sagittal left to right)|
|Slice thickness (mm)||2||1.4||2|
|Slice interval (mm)||2||0.7||1|
|Parallel imaging factor||-||2.5||3|
|Acquisition time||5 min 25 s||6 min 52 s||2 min 28 s|
TOF-MRV: Time-of-flight magnetic resonance venography, PC-MRV: Phase-contrast magnetic resonance venography, CE-MRV: Contrast-enhanced magnetic resonance venography, 2D: Two-dimensional, 3D: Three-dimensional, ms: Millisecond, mm: Millimeter, VENC: Velocity encoding gradient, cm/s: Centimeter per second, min: Minute, s: Second
POTENTIAL PITFALLS AND ARTIFACTS
A number of false positive and false negative errors are encountered during the MR evaluation of DST. For simplicity, we divide these pitfalls in the following three categories:
I. Errors on spin-echo sequences
Apparent “flow void” in acute DST
An acute thrombus (0–5 days) contains paramagnetic deoxyhemoglobin; hence, it appears isointense on T1W and hypointense on T2W images. Therefore, an acute thrombus may mimic a flow void seen in patent sinuses on SEI.[3,23] In any circumstance, luminal signals must always be evaluated on the gradient recalled echo (GRE) sequence/ susceptibility-weighted imaging (SWI) as well. These imaging sequences are very sensitive to field inhomogeneity caused by the paramagnetic acute thrombus, depicted as a hypointense signal (“blooming”/susceptibility artifact). The presence of susceptibility within dural sinuses must raise suspicion of DST and prompt confirmatory venography [Figure 4].[4,24,25]
Loss of flow void: Subacute DST versus slow flow
As the thrombus ages into the subacute phase (6-15 days), deoxyhemoglobin denatures into methemoglobin. This imparts a hyperintense signal on T1W and T2W images, an appearance commonly referred to as “loss of flow void.” At this stage, slow flow may mimic DST. With slow transit of venous blood, protons get rephased by the 180° pulse, rendering a bright intraluminal signal. Slow flow typically affects isolated segments and is generally isointense on T1W images.[3,14]
It is imperative to make the correct choice of MRV sequence in equivocal cases. TOF-MRV has its limitations in the diagnosis of subacute DST (discussed subsequently). PC-MRV, on the other hand, is insensitive to T1 characteristics of tissue and detects all flowing spins within a given VENC. Hence, optimization of the VENC (lower range) can reliably distinguish slow flow from subacute DST [Figure 5].[13,18,26]
In addition, evaluation of luminal signal on the GRE sequence is useful in these circumstances. Methemoglobin, being highly paramagnetic, causes “blooming” on GRE, which would not be expected in slow flow states.[3,14,24,25] A subacute thrombus often shows restricted diffusion reflecting its dense microstructure [Figure 6].
Entry slice phenomenon (ESP)
ESP refers to the paradoxical bright intraluminal signal observed on the initial slices of T1W sequences, raising concerns for subacute DST. This occurs due to the flow of unsaturated or “relaxed” spins into an imaging slice on the background of stationary spins saturated by the short TR. The signal intensity fades with successive slices as the blood progressively gets saturated by the rapid RF (short TR) [Figure 7].[13,14,28]
ESP is readily identified as it is limited to only the initial imaging slices [Figure 8]. If MRV is considered, PC-MRV is preferable as it allows visualization of flow independent of T1 signals. Another easy method to differentiate artifactual flow signal from a true thrombus is repeating acquisition in a different plane. Subacute thrombus shows a bright signal irrespective of orientation of the imaging plane, whereas ESP disappears [Figure 9].
II. Errors on MR venograms
“Thrombus shine through” effect
The short TR interval used in TOF-MRV makes it a surrogate T1W sequence. Tissues with native intravoxel T1 shortening tend to “shine through,” that is, they appear hyperintense on TOF exams. Thus, methemoglobin in subacute DST being T1 hyperintense contributes to an artifactual bright signal on TOF-MRV. Although this is usually not as bright as flow signals, it may still be misinterpreted as patency [Figure 10].[3,14] Careful scrutiny of GRE/SWI images (for blooming/susceptibility artifacts) and use of CE-MRV with subtraction help to confirm thrombosis [Figure 11]. PCMRV is very useful in such cases because it is insensitive to T1 shortening effects.[3,14,17,18]
In-plane saturation (IPS)
Flow signals on TOF images are brightest when blood enters perpendicular to the imaging plane. If dural sinuses lie parallel (“in-plane”) to the acquired sections, the spins get saturated and do not contribute signals. This phenomenon of signal loss, in an otherwise patent dural sinus, is known as “in-plane saturation” and must be distinguished from DST.[30,31]
Understanding the orientation of acquisition is key to identifying IPS. For example, obtaining coronal TOF venograms aids in the adequate evaluation of the midline sinuses; however, IPS may be troublesome at the torcular and transverse sinuses. For confirmation, orthogonal switching of acquisition may eliminate IPS related signal loss [Figure 12].[30,31]
In an attempt to eliminate acquisition of TOF-MRV in two perpendicular planes, a double oblique plane (45° to both coronal and sagittal planes) has been advocated; however, it must be noted that no rectilinear plane can be oriented perpendicular to all dural sinuses in a single acquisition. Reconstruction of an obliquely acquired dataset into standard maximum intensity projections also poses challenges due to software limitations.
Hypoplastic dural sinuses and flow gaps on TOF-MRV
TOF-MRV can detect slow flow, but only up to a minimum limit. Spins flowing slower than this limit encounter the short interval RF pulses and get saturated, resulting in signal loss, also known as “flow gap.” This phenomenon is further compounded by coplanar acquisition and is common in hypoplastic dural sinuses. In a study by Ayanzen et al., it was noted that flow gaps were present in 31% of non-dominant transverse sinuses. In fact, they did not observe flow gaps in dominant transverse sinuses or in the rest of the dural venous system, making it imperative to exclude thrombosis before concluding a flow gap. The lack of thrombus related signals and use of CE-MRV (independent of the flow dynamics) aid in confirming patency of sinuses [Figure 13].
Signal loss on PC-MRV
As discussed previously, PC-MRV displays flow information due to phase shifts in mobile protons within a velocity range (VENC) pre-selected by the MR operator. Flow velocity outside the selected VENC may not be displayed, leading to signal loss, mimicking an occlusion. This phenomenon is common in the lateral third of the transverse sinuses due to local flow turbulence caused by arachnoid granulations or stenosis in idiopathic intracranial hypertension (IIH) [Figure 14].
Fera et al. studied this signal loss in IIH and suggested that lower VENC values (15 cm/s) could not display high-velocity turbulent flow. The signal gap disappeared using a higher VENC (40 cm/s). Careful selection of the VENC is therefore vital in differentiating stenosis from a thrombus. CEMRV being independent of flow/geometry of sinuses and along with its high spatial resolution, is an essential adjunct to PC-MRV when the cause of the signal drop remains unclear.
III. Errors on CE MRI
With evolution into chronicity, DST demonstrates T1 isointense and T2 hyperintense signals. Pathologically, this phase is characterized by intrathrombus fibroblast and capillary proliferation. The end result is the conversion of the thrombus into a fibrotic mass with multiple endothelium-lined channels within. These channels enhance intensely, thereby rendering a false sense of normalcy on contrast-enhanced sequences. Furthermore, chronic DST may lack blooming/susceptibility artifacts (attributed to macrophages clearing denatured hemoglobin from the organizing thrombus), making the sequence unreliable [Figure 15]. Leach et al. reported susceptibility artifacts in 90.6% of DST in the initial 7 days as compared to 23.3% in DST older than 8 days. Unenhanced MRV is generally adequate to evaluate the lumen in this situation.[3,35]
Non-thrombotic filling defects
Arachnoid granulations are invaginations of arachnoid mater into venous sinuses through dural apertures. They serve as sites for CSF resorption. They can occur anywhere along the dural sinuses, especially along the lateral portion of the transverse sinuses (left predominance) followed by the superior sagittal sinus (SSS). On unenhanced T1W images, they appear isointense to hypointense and are always hyperintense on T2W images. Incomplete FLAIR suppression occurs due to altered CSF flow within. Arachnoid granulations vary in size, but when large, may mimic a thrombus [Figure 16].[3,36] In addition to their characteristic signal intensity, imaging features which help differentiate these filling defects from thrombosis include their focal, well-defined, round-oval, or lobulated shape and lack of susceptibility artifacts on GRE/ SWI sequences.[36,37] Arachnoid granulations are never hyperintense on T1W images. A frequent, though not consistent finding is the presence of cortical veins entering arachnoid granulations.[36,37] Liang et al., in a study of normal structures within dural sinuses, reported the presence of one or more veins in 414 out of 433 (96%) arachnoid granulations.
Fibrotic septations are normally found within dural sinuses, often in the straight and transverse sinuses oriented along their long axis. As compared to filling defects in partially recanalized DST, septations produce thin linear filling defects with sharp, well-defined margins. Occasionally, arachnoid granulations may be encountered at the end of septation.
Dural sinus stenting has been used for the treatment of IIH with great technical and clinical results. Transverse sinus stenting is known to improve vision, headache, and papilledema in patients with refractory IIH. These stents appear as filling defects on CE-MRV; they seldom pose concerns in the presence of an adequate clinical history [Figure 17].
MRI as a diagnostic tool has revolutionized the management of DST. DST can be diagnosed using standard MR sequences along with MRV even without contrast administration. A sound understanding of the principles of relevant MR techniques, awareness of imaging appearance of evolving thrombi, as well as potential imaging pitfalls are all crucial for avoiding misdiagnosis. We hope that these image-based diagnostic pearls come in handy, especially for trainee radiologists when dealing with these scans on a routine basis, so as to facilitate optimum management of this clinical scenario.
Declaration of patient consentPatient’s consent not required as patients identity is not disclosed or compromised.
Financial support and sponsorshipNil.
Conflicts of interestThere are no conflicts of interest.
- Am J Neuroradiol. 2007;28:940-5.Cerebral venous thrombus signal intensity and susceptibility effects on gradient recalled-echo MR imaging.
- [Google Scholar]
- MRI in Practice (4th ed). United States: Wiley-Blackwell Publishing Ltd; 2011. p. 262-89.Vascular and cardiac imaging. In:
- [Google Scholar]
- Am J Neuroradiol. 1994;15:1657-64.Time-of-flight intracranial MR venography: Evaluation of the sequential oblique section technique.
- [Google Scholar]
- Am J Neuroradiol. 2006;27:1913-8.2D time-of-flight MR venography in neonates: Anatomy and pitfalls.
- [Google Scholar]
- Am J Neuroradiol. 2000;21:74-8.Cerebral MR venography: Normal anatomy and potential diagnostic pitfalls.
- [Google Scholar]
- Am J Neuroradiol. 2007;28:782-9.Partially recanalized chronic dural sinus thrombosis: Findings on MR imaging, time-of-flight MR venography, and contrast-enhanced MR venography.
- [Google Scholar]
- Am J Neuroradiol. 2002;23:1739-46.Normal structures in the intracranial dural sinuses: Delineation with 3D contrast-enhanced magnetization prepared rapid acquisition gradient-echo imaging sequence.
- [Google Scholar]