Abstract
Objectives
To assess the diagnostic accuracy and lesion conspicuity of susceptibility-weighted angiography (SWAN) and T2* for the clot detection in acute cerebral venous thrombosis (CVT) by comparison with contrast-enhanced MR venography.
Methods
Venous thrombi detection and conspicuity were assessed by two readers for 18 venous segments on both T2*, SWAN source images, 2D SWAN reformats matching with T2*, and 3D SWAN images (SWAN-MinIP). Images obtained with the three reading techniques were systematically scored and compared to CE MRV findings, in a blinded fashion, per patient and per segment, and compared to each other.
Results
In 30 patients, 137 thrombosed venous segments were evaluated. The sensitivity of T2*, SWAN source images, 2D SWAN, and SWAN MinIP were, respectively, of 89.3%/82.1%, 82.1%, and 82.1% for dural sinus thrombosis and of 100%/100%/100%/96.6% for cortical venous thrombosis. There were significant differences in thrombus detection between T2* and SWAN: T2* versus SWAN source images and 2D SWAN (p = 0.04) and versus SWAN MinIP (p = 0.03). There were no significant differences between the three modalities of SWAN images. T2* was more sensitive than all SWAN images for both sigmoid sinus thrombosis and intracranial internal jugular vein thrombosis (p = 0.04). Inter-observer agreement was slightly superior with T2* (p < 0.05).
Conclusion
In this small cohort, SWAN sequence at 3T did not yield additional value for thrombus detection in acute CVT compared to T2*. This study highlights SWAN’s greatest weakness both for diagnostic accuracy and lesion conspicuity compared to T2* for acute venous clot detection near the skull base.
Keywords: GRE T2-T2*, susceptibility weighted angiography (SWAN) sequence, susceptibility-weighted imaging, cortical venous thrombosis, venous sinus thrombosis
Introduction
A lot of MR sequences have been proposed for the comprehensive examination of acute CVT, none of them being able to make an accurate diagnosis by itself.1-16 Contrast-enhanced MR venography has long been the reference standard.6,11,12
Susceptibility-weighted imaging is valuable for characterization of intracranial hemorrhage. T2-weighted Gradient-echo (T2 GRE, T2*) sequence is currently accepted as a standard MR sequence because of its sensitivity to acute intracranial hemorrhage. 17 It is also integrated into the MRI protocol investigating suspected cerebral venous thrombosis (CVT). On the other hand, susceptibility weighted angiography (SWAN) is a specific susceptibility-weighted imaging (SWI) sequence that was initially developed to assess small arterial involvement in the brain. Indeed, SWAN has shown a greater sensitivity compared to T2* in the detection of distally located intra-arterial thrombi in acute ischemic stroke.18-19 In a similar way, several studies confirmed the high sensitivity of susceptibility-weighted 3D-gradient-echo imaging technique for the detection of small cerebral veins, vessels rich in deoxygenated blood, small intracranial hemorrhages and microbleeds.20-24 Additionally, SWI has been used for the evaluation of venous stasis in CVT at 1.5 T 14 and for the evaluation of venous volume changes in patients with CVT at 3T. 25
On the other hand, T2* has been applied in CVT patients, showing a high sensitivity and additional value for the detection of acute venous clots.2,3,5,7 Optimizing the MRI protocol at the acute stage of CVT still remains a challenging issue. The diagnostic value in acute CVT of other MRI sequences with and without contrast such as contrast-enhanced (CE) T1, CE 3D T1- GRE, and 3D black-blood imaging have also been assessed and compared in several studies.6,11-13 A new technique—3D T2-SPACE, even if providing new additional data, has been proposed to be incorporated to comprehensive MR protocols. 26 However, SWAN sequence has never been compared with T2* or any other MR sequence, in particular with CE MR venography (CE MRV).11,12
The aim of this retrospective study was hence, regarding the detection of acute CVT to a) assess the diagnostic accuracy and lesion conspicuity of SWAN sequence with T2*, by comparison with CE MRV; b) evaluate whether SWAN can act as a surrogate to T2*.
Patients and methods
Patients’ selection
Inclusion criteria of the present study were as following: (1) CVT was considered acute if the duration of symptoms from onset to admission was of 0–7 days; (2) all the patients underwent our standardized MR brain protocol (T1, FLAIR, T2, DWI), including in addition for CVT diagnosis, both T2*, SWAN and CE MRV sequences.
Patients were excluded if all the sequences were not acquired or in case of poor imaging quality because of motion artifacts. For this study, informed patient consent was not required as patients’ medical records and imaging data were acquired as part of routine clinical care.
Venous thrombus location
Eighteen separate venous segments were evaluated for patency evaluation: superior sagittal sinus (SSS); transverse sinus (TS), right (RTS), left (LTS); sigmoid sinus right (RSiS) and left (LSiS); right and left intracranial internal jugular veins (IIJV;) straight sinus (SS); confluence of sinuses (Torcular: T); vein of Galen (VG); internal cerebral veins (ICV); Labbé vein (LV); frontal veins, right (RFV) and left (LFV); parietal veins, right (RPV) and left (LPV).
Image acquisition: MRI protocol (Table 1, Supplemental data)
The acquisition time for SWAN and T2* was similar (182 vs 196 s).
Susceptibility-weighted angiography source images (acquired with a resolution of 1.6 mm), free mode MPR and Minimum intensity projection (MinIP) of the SWAN sequence (SWAN MPR view MinIP = SWAN-MinIP reconstructed with 5 mm section thickness) were used by the readers. Furthermore, SWAN images were reformatted into 5-mm-thick sections, with 1 mm increment (2D SWAN). The slice positioning was used in the same plane and slices had the same thickness as T2* images, to match images from both sequences. The reformatted images encompassed the whole brain.
Image analysis
All images were analyzed using the Carestream PACS system/program (Carestream Health, Rochester, NY, USA) and were reviewed by) two radiologists (MB and JPL) with 20 years’ experience each, in stroke imaging during a separate preliminary session.
Each anonymized set of MR images (i.e., T2*, SWAN source images, 2D SWAN, and MinIP images) was randomly reviewed on separate sessions.
T2* and all SWAN reading techniques (i.e., source images, 2D SWAN, and MinIP) were evaluated separately for the presence of thrombi in 18 venous segments in each patient. The readers were blinded to clinical data and imaging data on which the diagnosis of CVT was based (i.e., CE MRV) and were blinded to each other. Disagreements between them were solved by consensus. To assess inter-observer variability, randomly selected sequences were independently analyzed by the two observers. A new reading session was conducted by one of us (MB) to assess intra-observer variability. For T2* and SWAN sequence, venous thrombus was defined as a characteristic hypointense signal within dural venous sinus (DVS) and veins. The two readers compared the two sequences for lesion detectability and conspicuity (image quality).
Thrombus detection
The diagnostic performance of T2* and all SWAN images for acute CVT diagnosis was evaluated per patient and per segment. For each segment, thrombus detection on T2* and all SWAN images was defined as the presence or absence of thrombus, according to a four-grade scale: 1: No thrombus, 2: ill-defined = low thrombus probability, 3: probable thrombus, 4: definite thrombus. The point scoring methodology used referred to similar study designs investigating venous thrombosis. 1 For sensitivity and specificity analysis, calculated per patient and per segment, scales three and four were both classified to be “thrombus.” Thrombus was classified per segment, according to the 16 venous segments described above. Thrombosed venous segments, possibly detected by susceptibility sequences (T2* and SWAN) and not detected by CE-MRV, were included in the statistical calculations using the kappa test.
Image quality
For image quality (lesion conspicuity), a four-grade scale was used according to presence or not of blurring and good or poor contrast: 1: non-diagnostic (Figure 1), 2: boundary of thrombosed venous segment not sharp, interfering with image interpretation (Figure 2), 3: boundary of thrombosed venous segment partially sharp with no clear-cut delineation, or minor blurring but sufficient for diagnosis (Figure 3), 4: boundary of thrombosed venous segments perfectly sharp (Figure 4).
Figure 1.
The missed thrombus. Contrast-enhanced MR venography (CE MRV) showed thrombosis of the left sigmoid sinus (SiS) and intracranial internal jugular vein (a). The venous thrombus was missed by the two readers using the T2* (b), 2D SWAN (c), SWAN MinIP reformations (d), and SWAN source images (e). To note: the patent right SiS has the same aspect than the thrombosed left SiS on SWAN images.
Figure 2.
The boundary of the thrombosed venous segments is not sharp, interfering with image interpretation in three patients: thrombosis of left sigmoid sinus on T2* (a) and 2D SWAN (b), of left transverse sinus thrombosis on SWAN MinIP (c), and of left sigmoid sinus thrombosis on SWAN-MinIP (d).
Figure 3.
The boundary of the thrombosed venous segments is partially sharp. The diagnosis is possible: in Sigmoid sinus thrombosis on T2* (a) and 2D-SWAN (b), and in transverse sinus thrombosis on 2D-SWAN (c), and of left sigmoid sinus thrombosis on SWAN-MinIP (d)
Figure 4.
The boundary of the thrombosed venous segment is perfectly sharp. The right transverse sinus thrombosis objective on contrast-enhanced MR venography (CE MRV) (a) is easily identified on both T2* (b), 2D SWAN (c), SWAN MinIP (d), and SWAN source image (e).
Comparison
All SWAN images were systematically compared to T2*images. A 3-grade classification was utilized to compare thrombus detection and lesion conspicuity as follows: 1: all SWAN images inferior to T2*, 2: all SWAN images equal to T2*, 3: all SWAN images superior to T2*. Additional findings (venous stasis, localized microhemorrhages) were recorded. Differential diagnoses as the presence of dural arteriovenous fistula were systematically searched for.
Statistical analysis
Statistical analysis was performed with SPSS 13.0 software for Windows (SPSS Inc., Chicago, USA). The quantitative values were presented as means ± standard deviations. Mc Nemar’s χ2 test was used to compare the detection rates of venous involvement between the three reading techniques. Inter- and intra-observer agreement was evaluated using the “k” statistics, with agreement considered as poor when “k” was <0.4, fair: 0.4–0.59, good: 0.6–0.74 and excellent: > 0.75.
The differences in the quantitative results between the three reading techniques were assessed using Wilcoxon signed-rank test. A p value < 0.05 was considered statistically significant.
Results
Between January 2012 and December 2018, a total of 30 patients (16 females; age range 18–80, mean [SD]: 44 [16.5], median: 45) presenting with acute CVT and matching the inclusion criteria were included. A total of 147 thrombosed venous segments were identified. Demographic and distribution of venous thrombosis are summarized in Table 1.
Table 1.
Demographics data and Thrombus location.
| Variables | Patients n = 30 |
|---|---|
| Demographic | |
| Female, n (%) | 17 (56.7) |
| Age (years), mean [SD], median | 44 [16.5], 45 |
| Thrombus localization, n segments | 137 |
| SSS | 16 |
| RTS | 14 |
| RSiS | 9 |
| LTS | 15 |
| LSiS | 10 |
| IIJV | 17 |
| SS (Straight sinus) | 4 |
| T (torcular) | 16 |
| RLV (right LabbéVein) | 2 |
| RFV | 7 |
| LFV | 10 |
| RPV | 5 |
| LPV | 10 |
| ICV | 2 |
SD: standard deviation; SSS = superior sagittal sinus; RTS = Right transverse sinus; RSiS = Right sigmoid sinus; LTS = Left transverse sinus; LSiS = Left sigmoid sinus; IIJV = intracranial internal jugular vein; SS = straight sinus; T = Torcular (confluence of sinuses); RFV, LFV: Right and Left frontal vein; RPV and LPV: Right and left parietal vein; ICV: Internal cerebral vein.
Thrombus detection
CE MRV showed thrombosis in 137 venous segments. Of these, 123 were identified on T2* (sensitivity (s) = 90.5%), 110 on SWAN source images and 2D SWAN (s = 80.3%), and 109 on SWAN MinIP (s = 79.5%) (Table 2). These results differed according to the involved venous sinus segment.
Table 2.
Diagnostic Performance per venous segment of T2*, SWAN source images, 2D SWAN and SWAN MinIP for detection of acute CVT.
| T2* | SWAN source images | 2D SWAN | SWAN MinIP | |
|---|---|---|---|---|
| n (%) | n (%) | n (%) | n (%) | |
| Thrombosed venous segments (137) | 123(90.5) | 110 (80,3) | 110 (80,3) | 109 (79,5) |
| Thrombosed dural venous sinuses (84) | 75 (89,3) | 69 (82,1) | 69 (82,1) | 69 (82,1) |
| SSS (16) - TS (29) - T (16) - SS (4) (total = 65) | 65 (100) | 65 (100) | 65 (100) | 65 (100) |
| SiS (19) | 11 (57,9) | 3 (17.6) | 3 (17.6) | 3 (17.6) |
| IIJV (17) | 10 (58.8) | 3 (17.6) | 3 (17.6) | 3 (17.6) |
| Cortical/anastomotic veins thrombosis (30) | 30 (100) | 30 (100) | 30 (100) | 29 (96.7) |
| Specificity (%) | 100 | 100 | 100 | 100 |
| Sensitivity (%) | 90.5 | 80.3 | 80.3 | 79.5 |
CVT = cerebral venous thrombosis; SSS = superior sagittal sinus; TS = transverse sinus; T = Torcular (confluence of sinuses); SS = straight sinus; SiS = sigmoid sinus; IIJV = intracranial internal jugular vein.
Indeed, considering all the venous sinus thrombosis (84/137 segments), the sensitivity of T2* and all SWAN images was 89.3% and 82.1%, respectively.
Considering all together SSS, TS, T and SS, the sensitivity was 100% for both T2* and all SWAN images, while for SiS and for IIJV thrombosis, the sensitivity was, respectively, of 57.9% and 58.8% for T2* and of 17.6% and 17.6% for SWAN images (Table 3).
Table 3.
Score for each sequences for detection and image quality.
| Detection | Image quality | |||||||
|---|---|---|---|---|---|---|---|---|
| T2* | SWAN source I | 2D SWAN | SWAN minIP | T2* | SWAN source I | 2D SWAN | SWAN minIP | |
| Mean [SD] scores | 3.44 [1.01] | 3.13 [1.19] | 3.08 [1.15] | 3.12 [1.14] | 3.39 [1.01] | 3.04 [1.14] | 3.08 [1.15] | 3.00 [1.11] |
| Min | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| Med (50th percentile) | 4 | 4 | 4 | 4 | 4 | 4 | 4 | 4 |
| Max | 4 | 4 | 4 | 4 | 4 | 4 | 4 | 4 |
| Normality test KS | 0.3999 | 0.4349 | 0.2983 | 0.3402 | 0.2568 | 0.3316 | 0.2770 | 0.3577 |
| Normality test p value | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
SWAN source I: Swan source image.
When considering all sequences, the most overlooked thrombosed venous segments were the SiS and the IIJV. Within them: T2* overlooked SiS thrombosis in 8/19 and IIJV thrombosis in 7/17, while both SWAN images overlooked these thrombosed segments in 16 and 14 cases, respectively, (Figure 1(b) and (c)).
Sensitivity and specificity for detection of CoVT (32/137 segments) was 100% for T2*, SWAN source images and 2D SWAN and 96.6% for SWAN MinIP. Image blurring for CoVT was observed on SWAN in one out of two cases of LV thrombosis (LVT) that presented with associated superficial temporal lobe hematoma (Figure 5). In this case, LVT well identified on T2*, source images and 2D SWAN was overlooked on SWAN MinIP because this sequence failed to distinguish the strong hypo-intensity of the thrombosed LV from the strong hyposignal of the temporal bone (Figure 5).
Figure 5.
Labbé vein thrombosis (LVT). T2* (a) and 2D SWAN images (b), and show an extra-axial dark band at the left temporal convexity, related to LVT. SWAN MinIP image fails to distinguish the strong hypointensity of the thrombosed LV from the strong hyposignal of the temporal bone (c).
Evaluating CVT, both T2* and SWAN images did not detect any additional findings, and any arterio-venous fistula. As SWAN sequence permits multiplanar reformations (oblique and/or curvilinear MPR), it allowed to edit trace, when necessary, of the entire thrombus length within a cortical vein (Figure 6(g)).
Figure 6.
Parietal vein thrombosis, not identified on 3D projection of contrast-enhanced MR venography (CE MRV) (a), is objectivated on multiplanar reformat (MPR) as a non-filling of its enlarged lumen with enhancement of the inflammatory venous wall (b) and on post-contrast 3D T1 reformat (c, arrow). The parietal vein thrombosis was detected by the two readers as a dark, curvilinear structure on T2*, 2D SWAN, SWAN MinIp and SWAN source images (d, e, f, and g). Curvilinear reformat from SWAN provides better information about the thrombus length within the parietal vein (h). T2* image shows small subdural effusions, not objectivated on SWAN, in a patient with intracranial hypotension. MPR from CE-MRV shows diffuse pachymeningeal enhancement (b).
Overall, T2* displayed higher sensitivity than SWAN in the depiction of venous thrombosis (T2* vs SWAN source images and 2D SWAN, p = 0.04; T2* vs MinIP, p = 0.03). Considering only the skull base venous structures, T2* images were more sensitive than all SWAN images for both SiS thrombosis and IIJV thrombosis (p = 0.04). None of the sequences at the segmental level displayed any false positive result, resulting in a specificity of 100%. Nonetheless, incomplete thrombus occlusion of the SSS (n = 4) and of the TS (n = 2) were identified on SWAN MinIP as complete sinus occlusion, while on SWAN source images and 2D SWAN an hyperintense signal within the sinus indicated a persistent flow (Figure 7).
Figure 7.
Incomplete occlusion of the Superior Sagittal Sinus (SSS) by the thrombus identified on contrast-enhanced MR venography (CE MRV) (a) is objectivated on T2* (b), SWAN 2D (c), and SWAN source image (d), as a hyper-intense area within the SSS—related to blood flow—and is not identified on SWAN MinIP reformat (e).
The overall scores of each image for detection are presented in Table 3.
This study shows a better thrombus detection rate of T2* compared to SWAN images:
T2* versus Swan MinIP (p = 0.02), vs 2D Swan (p = 0.03) and versus source images (p = 0.05). Al contrary there was no significant differences between the three modalities of SWAN images (Swan 2D vs Swan MinIP: p = 0.88; 2D vs source images, p = 0.91; MinIP vs source images, p = 0.81).
Analyzing the signal of the thrombus, we observed an iso-to-hyperintense signal on T1 and FLAIR in 52 and 40, respectively, and in 31 segments on DWI. In other words, at best, these sequences detected a thrombus in 37% of cases. We also observed that hyperintense signal (HIS) venous segments on DWI (n = 31, 22.8%) sequence were concomitantly associated with HIS on T1 and on FLAIR sequences. All these thrombi displayed a signal void on T2* and SWAN sequences.
In addition, thrombosis of seven cortical veins (isolated in 1 case) highlighted by susceptibility techniques was not visualized by the CE-MRV. Cortical veins are highly variable in number, size, and location, which make their diagnosis very challenging, which might be somewhat reliable for the underestimation of isolated CoVT. The kappa test was 0.87, indicating good agreement between susceptibility techniques and CE-MRV.
Image quality
Image quality (lesion conspicuity) was significantly different between T2* and SWAN sequences. The mean SWAN score (source images, 2D SWAN and SWAN MinIP) was inferior to the mean T2* score, overall and separately for venous sinus and cortical veins.
The overall scores of each image for detection are presented in Table 4.
The mean score differences for image quality between T2* and both SWAN images were statistically significant: T2* versus Swan MinIP (p = 0.003), vs 2D Swan (p = 0.03) and versus source images (p = 0.01). There were no significant differences between the different SWAN images: 2D versus MinIP: 0.45, 2D versus source images: 0.7 and MinIP versus source images: 0.72.
Venous stasis was observed in 8/26 cases, and better identified on SWAN images (both source images, 2D SWAN and MinIP) than on T2* images (p value non calculated due to the small number). Furthermore, venous stasis was better visualized using source images and SWAN MinIP than 2D SWAN (Figure 8).
Figure 8.
Venous stasis and microhemorrhages in left transverse sinus thrombosis. Venous stasis (prominence of cortical veins) over the temporal lobe, is identified on SWAN (a, stars) and not on T2* (b). 3-months-follow-up MRI shows disappearance of venous stasis on SWAN (c). Note that SWAN sequence detects more micro-hemorrhages and allows a better visualization (a, arrows) than T2* (b).
Localized microhemorrhages related to CVT were identified in 2 cases on both sequences although the number of detected lesions was higher and lesion conspicuity was better on SWAN (Figure 8). Conversely, no microbleeds and no dural arteriovenous fistula were detected by the three techniques.
Comparison between observers
Inter-observer agreement was better for T2* than 2D SWAN, MinIP SWAN, and source images (k = 0.87 vs 0.81, 0.79 and 0.79, p < 0.05). Intra-observer agreement was moderately higher for T2*, than 2D SWAN MinIP SWAN and source images (k = 0.94 vs 0.88, 0.82 and 0.83, respectively; p = NS). T2* was the most robust sequence with less inter and intra-observer reliability.
Discussion
At present, the diagnostic impact of SWAN in acute CVT detection has not yet been assessed nor its performance compared to that of T2* sequence. Our study shows that, at the acute stage of CVT, T2* exhibited a higher sensitivity than SWAN sequence and that this greater sensitivity affected mainly the skull base venous structures. In addition, SWAN was less sensitive to discriminate between complete and incomplete occlusion of a venous sinus by the thrombus.
In this study, there was a significant difference between SWAN (source images, 2D reformat SWAN and SWAN MinIP) and T2* images for acute CVT detection at the price of similar acquisition time. Because of its sensitivity to susceptibility changes, T2*is used to depict paramagnetic deoxyhemoglobin within the thrombus at the acute stage of CVT, resulting in marked signal loss. Susceptibility-weighted angiography sequence enhances susceptibility effect of deoxyhemoglobin, and hence provides significantly larger signal void of the thrombus than T2* sequence.
This strong susceptibility effect of SWAN generates a blooming effect affecting the edges of the thrombosed sinus that impairs either their delineation or their detection-even if their detection remains possible-depending on the location of the thrombosed sinus.
Our results show a sensitivity of 80.3% for both SWAN source images and 2D SWAN, of 79.5% for SWAN MinIP and of 90.5% for T2* images. Susceptibility-weighted angiography did not identify more thrombi thanT2*. However, these results should be balanced according to the thrombosed venous segment: the sensitivity of T2* and of three SWAN reading techniques was, respectively, of 88.4% and 78% for DVST and of 100%/100%/100%/95.2% for CoVT. But when considering SiS and IIJV thrombosis, the sensitivity dropped to 61.5% for T2* and between 15.4% and 16.6% for SWAN (3 reading techniques).
Detection of SSS thrombosis was easy on SWAN as on T2*. The blurring did not affect the delineation of the thrombus in this location, except in case of incomplete occlusion of the SSS by the thrombus.
In this occurrence, the persistent flow not identified on MinIP related to the blooming effect was depicted on source images and 2D SWAN as a bright signal within the SSS, proving incomplete venous occlusion by the thrombus. Our study showed that in thrombosis of the TS and of the sinus confluence (Torcular) the image quality was affected with SWAN (and more often on SWAN MinIP than on 2D SWAN), even if in these locations the thrombi were not overlooked.
The most striking finding was the non-detection of thrombus in the SiS and in the IIJV on both sequences, observed more often on SWAN than on T2* sequence. It is well-established that identification of small and distal thrombi within the cerebral arteries is facilitated by SWAN sequence, due to the overestimation of the size of the thrombi related to its strong susceptibility effect (blooming effect) compared to T2* sequence. Conversely, this phenomenon is disadvantageous for the delineation of venous thrombi near the skull base because the signal void is equivalent between bone and thrombosed sinus. 27 Consequently, the boundary of the thrombosed sinus in contact with bone was not seen or not sharp, or partially sharp. This phenomenon should be increased with the high magnetic field. The limitation of T2* assessment of SiS and IIJV thrombosis, because of susceptibility artifacts from the skull base, has been underlined by previous studies.1,6,7,11-13,28 Our findings suggest that the susceptibility artefacts from the skull base provide more significant limitations in SWAN, compared to T2* in the assessment of SiS and IIJV thrombosis.
On the other hand, for Boeckh-Behrens 27 SWAN sequence performed on a 3T system provides a better delineation of the patent venous system than T2*, and the SiS is not impaired by susceptibility artifacts from the skull base. Our findings are in contradiction with those of Boeckh-Behrens 27 : in the patent contralateral SiS and IIJV, the same signal void as in the thrombosed side was observed.
However, the blurring aspect in SWAN images was not found in cortical vein thrombosis, probably due to the smaller diameter of these venous structures and consequently of the thrombi. A recently published case report has shown that the thrombus within a DVA collector was delineated with the same accuracy by the two sequences. 29
To note, only occlusion of the largest veins is detectable on CE-MRV. 14
Each sequence provided advantages and drawbacks. First, SWAN was less accurate than T2* for the detection and delineation of the thrombi at the skull base. Second, SWAN (source images, 2D SWAN as SWAN MinIP) had a propensity to better depict vein congestion, thanks to a good visualization of the whole venous system, the thin imaging sections, and its high sensitivity to deoxyhemoglobin. In addition, SWAN showed more micro-hemorrhages related to CVT than T2* sequence.
These findings are in concordance with previous studies underlining the ability of SWI on 1.5 and 3T from other venders14,17 to evaluate venous stasis in CVT and the disappearance of cortical veins dilatation at the time of recanalization. 14 In addition, SWAN permitted curvilinear and MPR images, which could be useful in case of cortical veins thrombosis to evaluate thrombus length but has the disadvantage to be time-consuming. The high sensitivity of these sequences may be of value when CE-MRV is not available or contra-indicated.
Hypoplasia or atresia of TS—although not present in our series—is not an exceptional occurrence, and can lead to false thrombosis diagnosis. It is established that using both CE-MRV and non-contrast MRV (i.e., phase-contrast MRV and Inhance 3D Velocity useful when the use of gadolinium contrast is contraindicated), unilateral decreased signal/non-visualization of the TS do not allow to differentiate them from TS thrombosis.16-29 Recently, Zhu et al. 30 reported that Time-Resolved MRA (TR-MRA) had higher accuracy in distinguishing TS thrombosis from TS hypoplasia.
Both T2*and SWAN sequences, allowing direct visualization of thrombus as opposed to indirect detection of obstacle to venous flow, have their place in the diagnosis of acute CVT; they are further the only techniques to not overlook a congenitally atretic/hypoplastic venous segment.
The differences between our findings and literature reports regarding the T1, FLAIR, and DW signal of thrombosed segments can be explained by the fact that on the one hand, these reports included both acute and subacute CVTs6,9,11,12,15,29,31 while our cohort only includes acute CVTs-and on the other hand that most of the studies focused on dural sinuses thrombosis,6,11,12,15,30 while our series consider all venous segments.
The signal intensity of venous thrombus, whatever the sequence, depends on the timing of the MRI compared to the onset of clinical signs and symptoms.
Both T2 * and SWAN have an added value for the detection of CoVT, as well as for not misdiagnose a hypoplasia/aplasia of the TS.
Nonetheless, as T2* has a higher performance than SWAN sequence at 3T for the detection of skull base venous thrombosis, T2* must be included in the comprehensive multiparametric MR imaging protocol for the diagnosis of acute CVT, along with CE-MRV. For accurate diagnosis at the acute phase, recognition of the pitfalls and limits of each sequence is essential.11,12,32
This study has several limitations. The relatively small number of included patients is the major limitation for the general conclusions made, even if this number reflects a very rare disease. This also explains the retrospective character of the study. The importance of the strong blooming effect may be related to the high field (3T) used in this study. The echo times of the two sequences were substantially different: TE of 2D T2* was 15 ms and TE of SWAN was 24 ms. The longer echo time may lead to much more signal loss and may be a major factor for the preference of 2D T2* in some cases, even if those different long TEs in both sequences are prone to similar susceptibility artifacts. The diagnostic value of SWAN sequence in acute CVT remains to be proven by larger cohorts, and with other devices including 1.5 T MR systems.
Conclusion
The current findings based on a short series of patients examined at 3T show that SWAN sequence did not detect a higher number of thrombosed venous segments in acute CVT than T2* sequence. Even if SWAN is superior in detecting venous stasis and brain micro-hemorrhages, this study highlights SWAN’s greater weaknesses both for diagnostic accuracy and image quality compared to T2* in acute SiS and IIJV thrombosis, due to the susceptibility artefacts.
Supplemental Material
Supplemental Material for Comparison of susceptibility-weighted angiography (SWAN) and T2 gradient-echo sequences for the detection of acute cerebral venous thrombosis by Monique Boukobza, Sylvain Guinebert and Jean-Pierre Laissy in The Neuroradiology Journal
Appendix.
Abbreviations
- CVT
Cerebral venous thrombosis
- 3D T1-weighted GRE CE
Contrast-enhanced 3D GRE T1-weighted sequence
- CE MRV
Contrast-enhanced MR venography
- CoV
Cortical veins
- DVS
Dural venous sinuses
- DVST
Dural venous sinus thrombosis
- GRE T2
T2-weighted gradient recalled echo sequence.
- MinIP
Minimum intensity projection
- MPR
Multiplanar reformation
- SWAN
Susceptibility-weighted angiography
- SWAN-MinIP
SWAN-Multiplanar Reformatted view MinIP
- SWI
Susceptibility-weighted imaging
- TS
Transvers sinus
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article.
Ethical standards: No institutional review approval was required by the Ethics Committee of our institution for retrospective analysis of patients’ medical records and imaging data acquired as part of routine clinical care. Specific national laws have been observed.
Supplemental Material: Supplemental material for this article is available online.
ORCID iD
Monique Boukobza https://orcid.org/0000-0002-8691-9004
References
- 1.Liang L, Korogi Y, Sugahara T, et al. Evaluation of the intracranial dural sinuses with a 3D contrast-enhanced MP-RAGE sequence: prospective comparison with 2D-TOF MR venography and digital subtraction angiography. AJNR Am J Neuroradiol 2001; 22(3): 481–492. PMID: 11237970. [PMC free article] [PubMed] [Google Scholar]
- 2.Idbaih A, Boukobza M, Crassard I, et al. MRI of clot in cerebral venous thrombosis: high diagnostic value of susceptibility- weighted images. Stroke 2006; 37(4): 991–995. DOI: 10.1161/01.STR.0000206282.85610.ae [DOI] [PubMed] [Google Scholar]
- 3.Boukobza M, Crassard I, Bousser MG, et al. MR imaging features of isolated cortical vein thrombosis: diagnosis and follow-up. AJNR Am J Neuroradiol 2009; 30(2): 344–348. DOI: 10.3174/ajnr.A1332 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Linn J, Michl S, Katja B, et al. Cortical vein thrombosis: the diagnostic value of different imaging modalities. Neuroradiology 2010; 52(10): 899–911. DOI: 10.1007/s00234-010-0654-0 [DOI] [PubMed] [Google Scholar]
- 5.Ihn YK, Jung WS, Hwang SS. The value of T2*-weighted gradient-echo MRI for the diagnosis of cerebral venous sinus thrombosis. Clin Imaging 2013; 37(3): 446–450. DOI: 10.1016/j.clinimag.2012.09.003 [DOI] [PubMed] [Google Scholar]
- 6.Saindane AM, Mitchell BC, Kang J, et al. Performance of spin-echo and gradient-echo T1-weighted sequences for evaluation of dural venous sinus thrombosis and stenosis. AJR Am J Roentgenol 2013; 201(1): 162–169. DOI: 10.2214/AJR.12.9095 [DOI] [PubMed] [Google Scholar]
- 7.Altinkaya N, Demir S, Alkan O, et al. Diagnostic value of T2*-weighted gradient-echo MRI for segmental evaluation in cerebral venous sinus thrombosis. Clin Imaging 2015; 39(1): 15–19. DOI: 10.1016/j.clinimag.2014.07.002 [DOI] [PubMed] [Google Scholar]
- 8.Sari S, Verim S, Hamcan S, et al. MRI diagnosis of dural sinus: cortical venous thrombosis: Immediate post-contrast 3D GRE T1-weighted imaging versus unenhanced MR venography and conventional MR sequences. Clin Neurol Neurosurg 2015; 134: 44–54. [DOI] [PubMed] [Google Scholar]
- 9.Yildiz ME, Ozcan UA, Turk A, et al. Diffusion-weighted MR imaging findings of cortical vein thrombosis at 3 T. Clin Neuroradiol 2015; 25: 249–256. [DOI] [PubMed] [Google Scholar]
- 10.Jalli R, Zarei F, Farahangiz S, et al. The sensitivity, specificity, and accuracy of contrast-enhanced T1-weighted image, T2*-weighted image, and magnetic resonance venography in diagnosis of cerebral venous sinus thrombosis. J Stroke Cerebrovasc Dis 2016; 25(8): 2083–2086. DOI: 10.1016/j.jstrokecerebrovasdis.2016.01.039 [DOI] [PubMed] [Google Scholar]
- 11.Patel D, Machnowska M, Symons S, et al. Diagnostic performance of routine brain mri sequences for dural venous sinus thrombosis. AJNR Am J Neuroradiol 2016; 37(11): 2026–2032. DOI: 10.3174/ajnr.A4843 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sadigh G, Mullins ME, Saindane AM. Diagnostic performance of MRI sequences for evaluation of dural venous sinus thrombosis. AJR Am J Roentgenol 2016; 206(6): 1298–1306. DOI: 10.2214/AJR.15.15719 [DOI] [PubMed] [Google Scholar]
- 13.Yang Q, Duan J, Fan Z, et al. Early detection and quantification of cerebral venous thrombosis by magnetic resonance black-blood thrombus imaging. Stroke 2016; 47(2): 404–409. DOI: 10.1161/STROKEAHA.115.011369 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Sato T, Sakuta K, Komatsu T, et al. Yield of combined MRI sequences in isolated cortical vein thrombosis diagnosis. J Neurol Sci 2017; 381: 328–330. DOI: 10.1016/j.jns.2017.09.008. [DOI] [PubMed] [Google Scholar]
- 15.Lv B, Tian CL, Cao XY, et al. Role of diffusion-weighted imaging in the diagnosis of cerebral venous thrombosis. J Int Med Res 2020; 48(6): 300060520933448. DOI: 10.1177/0300060520933448 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ozpar R, Tonkaz M, Erkal D, et al. Non-contrast magnetic resonance venography with Inhance 3D Velocity: diagnostic performance for intracranial venous thrombosis. Neuroradiology 2021; 63(11): 1853–1861. DOI: 10.1007/s00234-021-02710-1 [DOI] [PubMed] [Google Scholar]
- 17.Dempfle AK, Harloff A, Schuchardt F, et al. Longitudinal volume quantification of deep medullary veins in patients with cerebral venous sinus thrombosis: venous volume assessment in cerebral venous sinus thrombosis using SWI. Clin Neuroradiol 2018; 28(4): 493–499. DOI: 10.1007/s00062-017-0602-z [DOI] [PubMed] [Google Scholar]
- 18.Radbruch A, Mucke J, Schweser F, et al. Comparison of susceptibility weighted imaging and TOF-angiography for the detection of thrombi in acute stroke. PLoS One 2013; 8(5): e63459. DOI: 10.1371/journal.pone.0063459 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hodel J, Leclerc X, Khaled W, et al. Comparison of 3D multi-echo gradient-echo and 2D T2* MR sequences for the detection of arterial thrombus in patients with acute stroke. Eur Radiol 2014; 24(3): 762–769. DOI: 10.1007/s00330-013-3061-1 [DOI] [PubMed] [Google Scholar]
- 20.Cheng AL, Batool S, McCreary CR, et al. Susceptibility-weighted imaging is more reliable than T2*-weighted gradient recalled echo MRI for detecting microbleeds. Stroke 2013; 44(10): 2782–2786. DOI: 10.1161/STROKEAHA.113.002267 [DOI] [PubMed] [Google Scholar]
- 21.Hayashida Y, Kakeda S, Hiai Y, et al. Diagnosis of intracranial hemorrhagic lesions: comparison between 3D-SWAN (3D T2*- weighted imaging with multi-echo acquisition) and 2D-T2*-weighted imaging. Acta Radiol 2014; 55(2): 201–207. DOI: 10.1177/0284185113495836 [DOI] [PubMed] [Google Scholar]
- 22.Mahvash M, Pechlivanis I, Charalampaki P, et al. Visualization of small veins with susceptibility-weighted imaging for stereotactic trajectory planning in deep brain stimulation. Clin Neurol Neurosurg 2014; 124: 151–155. DOI: 10.1016/j.clineuro.2014.06.041 [DOI] [PubMed] [Google Scholar]
- 23.Shams S, Martola J, Cavallin L, et al. SWI or T2*: which MRI sequence to use in the detection of cerebral microbleeds? The Karolinska Imaging Dementia Study. AJNR Am J Neuroradiol 2015; 36(6): 1089–1095. DOI: 10.3174/ajnr.A4248 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Young A, Poretti A, Bosemani T, et al. Sensitivity of susceptibility-weighted imaging in detecting developmental venous anomalies and associated cavernomas and microhemorrhages in children. Neuroradiology 2017; 59(8): 797–802. DOI: 10.1007/s00234-017-1867-2 [DOI] [PubMed] [Google Scholar]
- 25.Hakim A, Kurmann C, Pospieszny K, et al. Diagnostic Accuracy of High-Resolution 3D T2-SPACE in Detecting Cerebral Venous Sinus Thrombosis. AJNR Am J Neuroradiol 2022; 26: 881–886. Epub ahead of print. DOI: 10.3174/ajnr.A7530 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Boeckh-Behrens T, Lutz J, Lummel N, et al. Susceptibility weighted angiography (SWAN)of cerebral veins and arteries compared to TOF-MRA. Eur J Radiol 2012; 81(6): 1238–1245. DOI: 10.1016/j.ejrad.2011.02.057 [DOI] [PubMed] [Google Scholar]
- 27.Leach JL, Wolujewicz M, Strub WM. Partially recanalized chronic dural sinus thrombosis: findings on MR imaging, time-of-flight MR venography, and contrast-enhanced MR venography. AJNR Am J Neuroradiol 2007; 28(4): 782–789. PMID: 17416839. [PMC free article] [PubMed] [Google Scholar]
- 28.Monique B, Sandra C, Marie-Cécile HF, et al. Serial imaging and SWAN sequence of developmental venous anomaly thrombosis with hematoma: Diagnosis and follow-up. J Clin Neurosci 2017; 43: 148–151. DOI: 10.1016/j.jocn.2017.05.002 [DOI] [PubMed] [Google Scholar]
- 29.Chang YM, Kuhn AL, Porbandarwala N, et al. Unilateral Nonvisualization of a Transverse Dural Sinus on Phase-Contrast MRV: Frequency and Differentiation from Sinus Thrombosis on Noncontrast MRI. AJNR Am J Neuroradiol 2020; 41(1): 115–121. DOI: 10.3174/ajnr.A6337 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Zhu Y, Wang D, Xie Y, et al. Differentiation of transverse sinus thrombosis from congenitally atretic transverse sinus with time-resolved contrast-enhanced magnetic resonance angiography. Clin Radiol 2022; 77(5): e372–e378. DOI: 10.1016/j.crad.2022.02.003 [DOI] [PubMed] [Google Scholar]
- 31.Favrole P, Guichard JP, Crassard I, et al. Diffusion-weighted imaging of intravascular clots in cerebral venous thrombosis. Stroke 2004; 35(1): 99–103. DOI: 10.1161/01.STR.0000106483.41458 [DOI] [PubMed] [Google Scholar]
- 32.Dmytriw AA, Song JSA, Yu E, et al. Cerebral venous thrombosis: state of the art diagnosis and management. Neuroradiology 2018; 60(7): 669–685. DOI: 10.1007/s00234-018-2032-2 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Material for Comparison of susceptibility-weighted angiography (SWAN) and T2 gradient-echo sequences for the detection of acute cerebral venous thrombosis by Monique Boukobza, Sylvain Guinebert and Jean-Pierre Laissy in The Neuroradiology Journal