'Susceptibility sign' on susceptibility-weighted imaging in acute ischemic stroke
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.96389
Source of Support: None, Conflict of Interest: None
Background and Aim: Acute intra-arterial thrombus produces susceptibility artifact on gradient echo images (susceptibility sign). Our aim was to study the sensitivity and specificity of the susceptibility sign in various major intracranial arteries on susceptibility-weighted imaging (SWI) in patients with acute stroke. We also compared it with the 'hyperintensity sign' on fluid-attenuated inversion recovery (FLAIR) and 'hyperdense artery sign' on computed tomography (CT) for middle cerebral artery (MCA) occlusion. Materials and Methods : We have retrospectively studied 48 patients with ischemic stroke in various stages, due to intracranial arterial occlusions, for presence of 'hyperdense artery sign' (CT), 'hyperintense arterial sign' (FLAIR sequence) and 'susceptibility sign' (SWI). The sensitivity and specificity of each sign to detect intracranial arterial occlusion were calculated using the contrast-enhanced magnetic resonance (MR) angiogram as reference standard. Results: The sensitivity and specificity of the 'susceptibility sign' for detecting the MCA occlusion were 77% and 100% respectively (10 of 13).The sensitivity of the 'susceptibility sign' for detecting anterior cerebral artery (ACA) occlusion was 50% (1 of 2), 66.6% for posterior cerebral artery (2 of 3) and 75% for basilar artery (3 of 4). All the vertebral artery occlusions showed 'susceptibility sign' (6 of 6). Overall sensitivity and specificity of the 'susceptibility sign' for all acute major intracranial arterial occlusions were 82% and 100% respectively. Only one of the two cases of subacute infarcts studied showed a positive susceptibility sign. None of the 11 chronic intracranial occlusions and seven internal carotid occlusions showed the sign intracranially. Conclusion: 'Susceptibility sign' is more sensitive in detecting the acute MCA thrombus as compared to 'hyperdense MCA sign' on CT and 'hyperintense artery' sign on FLAIR images. It also has high sensitivity and specificity for other intracranial acute arterial occlusions.
Keywords: Hyperdense middle cerebral artery sign, stroke, susceptibility sign
Intra-arterial thrombus can be identified on susceptibility-weighted images. Acute intra-arterial thrombus is paramagnetic due to the presence of a higher concentration of deoxyhemoglobin and clot retraction, which produces blooming artifact.  'Susceptibility sign' is said to be positive when the diameter of a hypointense vessel exceeds the diameter of the contralateral artery on susceptibility-weighted imaging (SWI) images.  In the current text, we use SWI to refer to magnitude or phase images, or a combination of both, obtained with a three-dimensional, fully velocity-compensated, gradient echo sequence. 'Hyperdense artery sign' is well-described in middle cerebral artery (MCA) territory and is less frequently seen in other intracranial locations. The 'hyperintense artery sign' on fluid-attenuated inversion recovery (FLAIR) images is observed only in MCAs.  Unlike this, the 'susceptibility sign' can be observed in all major intracranial arteries [Figure 1], [Figure 2] and [Figure 3]. Thus it adds valuable information in the evaluation of stroke patients. This article aims to compare 'susceptibility sign' (SWI), 'hyperdense artery sign'(computed tomography (CT)) and 'hyperintense artery sign' (FLAIR) for not only acute MCA occlusion, but also to evaluate the utility of this sign for detecting acute occlusions of the vertebral artery, basilar artery, posterior cerebral artery (PCA) and anterior cerebral artery (ACA) in patients with ischemic stroke.
We retrospectively analyzed the CT and magnetic resonance imaging (MRI) of 48 consecutive patients (35 males, 13 females with age ranging from 52 years to 80 years) done between January 2007 to December 2009 who presented with ischemic stroke and had intracranial arterial occlusions. Radiological investigation was performed in these patients for evaluation of acute (symptomatic for less than 24 h), subacute (symptomatic for 24 h to four weeks) or chronic stroke (symptomatic for more than four weeks). All the studies were done out of therapeutic window for anterior circulation stroke and none of the patients with either anterior or posterior circulation strokes underwent intra-arterial thrombolysis.
In all patients CT and MRI studies done within a 24-h interval of each other were available for review. All CT scans were performed on a single-slice GE scanner (Hi-Speed, Single-slice, GE Medical Systems, Milwaukee, US) with 5 mm section thickness. All MRI studies were obtained on a 1.5 Tesla scanner (Avanto Tim SQ engine, Siemens, Erlangen, Germany). The pulse sequences and parameters were as follows: (1) Diffusion-weighted image (DWI): Echo planar imaging (EPI) spin echo; Time to Repeat (TR): 3500 ms; Time to echo (TE): 105 ms; slice thickness: 5 mm; matrix size: 230×230; FOV: 230 mm; b values 0 and 1000 s/mm 2 (2) FLAIR: TR: 8000 ms; TE: 108 ms; slice thickness: 5 mm; matrix size: 160×256; FOV: 230 mm; Inversion time (TI): 2,500 ms; flip angle: 150° (3) SWI: TR: 49 ms; TE: 40 ms; slice thickness: 2.1 mm; matrix size: 512×256; FOV: 220 mm; flip angle: 20°; bandwidth, 80 kHz, acquisition time: 2.58 min (4) Magnetic resonance angiography (MRA): TR: 3.5 ms; TE: 1.1 min; slice thickness: 1.0 mm; matrix size: 192×256; FOV: 270 mm; flip angle: 30°. (5) Time-of-flight angiography (TOF MRA): TR: 36 ms; TE: 6.9 min; slice thickness: 1.4 mm; matrix size: 224×256; FOV: 18 cm; flip angle: 25°.
Two neuroradiologists independently analyzed two sets of images. The first set of images contained contrast-enhanced MR angiography (CE MRA) and time of flight (TOF) angiography, which was analyzed by the first radiologist for presence of intra-arterial occlusion in various major intracranial branches like anterior cerebral artery, middle cerebral artery, posterior cerebral artery, basilar artery and vertebral artery. The second neuroradiologist, who was blinded to MRA findings, was asked to determine the presence of the various signs described earlier ('susceptibility sign' on SWI, 'hyperdense artery sign' on CT and 'hyperintense arteries sign' on FLAIR) in acute (less than 24 h), subacute (between 24 h to four weeks) or chronic infarcts (more than four weeks), in different intracranial arteries. Age of infarcts was determined on the basis of clinical data and their appearances on diffusion-weighted, T2-weighted, FLAIR and T1-weighted images. Patients with internal carotid artery (ICA) occlusion not extending beyond the circle of Willis were separately considered. This was done because the susceptibility effect from the intracranial ICA occlusion is difficult to differentiate from artifacts due to the base of the skull. However, these images were analyzed for the presence of susceptibility sign on SWI images in intracranial arteries (distal to occlusion). These cases were included in this study to assess the effect of slow flow distal to the occlusion on SWI images.
Statistical analysis was done to evaluate the sensitivity and specificity of the susceptibility sign, hyperintense artery sign on FLAIR and hyperdense artery sign on CT for intracranial occlusion, considering the MRA as gold standard investigation. The positive predictive value (PPV) and negative predictive value (NPV) were also calculated for all the signs. Contralateral normal vessels of the same patient served as control for statistical calculation.
The results are summarized in [Table 1] and [Table 2]. Out of 48 cases, seven had isolated ICA occlusions not extending into the MCA or ACA and were excluded from the statistical analysis. Findings in the remainder of the 41 cases are summarized in [Table 1]. Out of 13 cases of acute MCA occlusions, 'susceptibility sign' was detected in 10 cases (sensitivity 77% and specificity 100%). 'Susceptibility sign' was also detected in one out of two cases with ACA occlusion, two out of three cases affecting the PCA, three out of four cases in basilar artery and all (six out of six) cases with vertebral artery occlusions. The sensitivity of 'susceptibility sign' for all acute intracranial arterial occlusions was 82% and the specificity was 100% (23/28). Only one case with subacute occlusion (one out of two cases) showed susceptibility sign. None of the cases with chronic occlusion showed positive susceptibility sign [Table 1].
Dense MCA sign was present in six out of 13 patients with acute MCA occlusions. Dense artery sign was also seen in one out of three cases in PCA, one out of four cases in the basilar artery and four out of six cases of vertebral artery acute occlusions. None of the chronic occlusions showed hyperdense artery on CT images. FLAIR showed 'hyperintense artery' in seven out of 13 patients with acute MCA occlusion.
Seven cases of isolated intracranial ICA occlusions which were not included in the above statistical calculations were evaluated separately for the presence of susceptibility sign. We did not appreciate any intracranial susceptibility sign in the circle of Willis branches on SWI images in these cases. In all these cases distal ACA and MCA branches were seen to reform through communicating arteries.
Magnetic properties of hemoglobin vary according to its oxygenation status. Oxyhemoglobin is diamagnetic and deoxyhemoglobin is a paramagnetic substance.  Appearance of blood on susceptibility-weighted images depends on the relative concentration of oxy- and deoxyhemoglobin.  The arterial blood has high T2 relaxation times (approximately 200 ms) as compared to venous blood (approximately 100 ms) at 1.5 T. , The intra-arterial thrombus contains more deoxyhaemoglobin as compared to flowing blood. Susceptibility sign occurs not only due to the increased paramagnetic effect from deoxyhemoglobin but also due to the increased hematocrit secondary to clot retraction and fibrin formation.  Not all thrombi produce blooming, instead there may be subtle changes in the paramagnetic effect as compared to rest of the flowing blood depending on content of the clot (whether white or red blood cells predominate in the white thrombus or red thrombus respectively). 
The SWI sequence is highly sensitive to the presence of paramagnetic substances.  SWI is composed of both, high-resolution magnitude and phase images. The phase mask images are derived from phase images and are multiplied with original magnitude images four times to yield images that improve the visibility of the susceptibility effect. , Final interpretation is done using these three sets of images. The magnitude images reveal the areas of short T2* signal losses but with less blooming effect due to high resolution. In phase images, paramagnetic substances appear hypointense, whereas diamagnetic substances appear bright (right-handed system). The final SWI processed images increase the visibility of the T2* signal intensity losses. ,
In our series, the detection of the 'susceptibility sign' was more sensitive than 'hyperdense MCA sign' detected on CT. It should also be noted that we used SWI sequence instead of a two-dimensional gradient recalled echo sequence (2D GRE), for detecting this. SWI, with thin sections and advanced phase-processing techniques, has been proven to be superior to 2D GRE to detect susceptibility changes.  Unlike the conventional T2* gradient images, SWI is a three-dimensional, velocity-compensated sequence that combines both magnitude and phase information to accentuate the susceptibility effect. 'Susceptibility sign' was initially described on conventional gradient images in MCA occlusions. Further studies are required to compare the susceptibility sign on conventional gradient echo sequences and on SWI.
In this series, we not only showed the utility of the 'susceptibility sign' in MCA occlusions but also in ACA, PCA, basilar and intracranial vertebral artery occlusions [Figure 1] and [Figure 3]. Interestingly, all the six acute vertebral artery occlusions showed a positive susceptibility sign, whereas only four cases showed hyperdensity on CT. 'Susceptibility sign' was seen in 75% (three out of four cases) of basilar artery occlusions whereas the CT showed unequivocal hyperdensity only in 25% (one out of four) cases. Even in PCA and ACA territories, performance of the susceptibility sign was better compared to the hyperdense artery sign [Table 2]. None of the chronic occlusions showed 'hyperdense artery sign'. We believe that thinner slices and higher in-plane resolution of the SWI images may have contributed to the observed higher sensitivity when compared to the thicker slices of FLAIR images and CT. Our findings agree with the previous studies which have shown that 'susceptibility sign' is more sensitive and specific than dense MCA sign on CT scan. , Assouline et al., reported 88% sensitivity and 100% specificity for MCA and PCA acute occlusion using the gradient echo sequences. 
Interestingly, only 7 out of 13 cases of MCA occlusion showed hyperintense arteries on FLAIR, whereas an additional three cases of MCA occlusion were detected by SWI. We found 'susceptibly sign' superior to FLAIR in overall detection of intracranial occlusions. This can be explained by the fact that the FLAIR hyperintensity sign is due to slow flow and retrograde collateral flow through leptomeningeal vessels, whereas the susceptibility artifact is detected directly due to the presence of intra-arterial thrombus. , This is also supported by the finding that all seven cases of ICA occlusions with no intracranial extension of thrombus did not show the susceptibility sign in MCA or ACA.
Hyperintense signal in FLAIR corresponds to perfusion abnormality and it is usually larger than the area seen on diffusion-weighted images. , The FLAIR sequence is also able to show the abnormal intra-arterial signals within 35 min of onset of symptoms with sensitivity equal to TOF angiogram. 
Advantage of the 'susceptibility sign' over the FLAIR hyperintense arterial sign is that it is not only seen in the proximal MCA but also in vertebral, basilar PCA and anterior cerebral arteries. In this study susceptibility artifacts were seen in 23 out of 28 cases in all locations (sensitivity 82% and specificity 100%). For MCA, sensitivity of 'susceptibility sign' on SWI is 10 out of 13 (sensitivity 77% and specificity 100%). All six cases of vertebral artery occlusion showed the susceptibility sign. None of the 10 cases with a chronic thrombus showed the susceptibility sign. Only one sub-acute thrombus (48 h after the onset of symptom) showed the persistent susceptibility sign (50%).
The main limitations of this study included bias due to retrospective data analysis and limited number of cases included. None of the cases were hyperacute and we do not know whether the susceptibility sign would have been detected on SWI within 6 h of the onset of ischemic stroke. The follow-up MRI images were not available for most of the cases of the acute infarct, to study the temporal evolution of the 'susceptibility sign'. In addition contrast-enhanced MRA was considered gold standard in our study instead of catheter angiogram. Further studies are required to evaluate the inter-observer agreement of susceptibility sign.
In conclusion, the susceptibility sign is a very useful sign in patients with acute stroke, especially with the use of the relatively new SWI sequence. It points to the location of the intra-arterial thrombus. Our study suggests that the 'susceptibility sign' has a high sensitivity for detecting acute intracranial arterial occlusion in all major intracranial arteries including anterior cerebral, middle cerebral, posterior cerebral, basilar and vertebral arteries while it was not visualized in patients with chronic arterial occlusion.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2]