Neurology India Medknow Publications on behalf of the Neurological Society of India ISSN: 0028-3886 EISSN: 1998-4022 Vol. 58, Num. 4, 2010, pp. 608-614

Neurology India, Vol. 58, No. 4, July-August, 2010, pp. 608-614

Topic of the Issue: Original Article

Susceptibility-weighted imaging in the evaluation of brain arteriovenous malformations

George Uttam, Jolappara Milan, Kesavadas Chandrasekharan, Gupta ArunKumar

Department of Imaging Sciences & Interventional Radiology, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Trivandrum - 695 011

Correspondence Address:Department of Imaging Sciences and Interventional Radiology, Sree Chitra Tirunal Institute for Medical sciences and Technology, Trivandrum - 695 011, chandkesav@yahoo.com

Date of Acceptance: 11-Jun-2010

Code Number: ni10160

PMID: 20739804

DOI: 10.4103/0028-3886.68668

Abstract

Keywords: Arteriovenous malformations, draining vein, feeding artery, nidus, susceptibility-weighted imaging

Introduction

Arteriovenous malformations (AVMs) are composed of a network of channels interposed between feeding arteries and draining veins, without any direct shunt. ^[1] Therapeutic approaches to AVMs include surgery, embolization, and radiotherapy. ^[2],[3] Identification and exact definition of the different vascular components are essential before planning the treatment. Digital subtraction angiography (DSA) remains the gold standard in the evaluation of AVMs. ^[4] In the past, magnetic resonance imaging (MRI) techniques, including time of flight (TOF) MR-angiography (MRA) and contrast-enhanced MR-angiography have been used in the noninvasive evaluation of AVMs.

Susceptibility-weighted imaging (SWI), a relatively new MRI technique, exploits the magnetic susceptibility differences of various tissues, such as blood, iron and calcification. ^[5] It consists of using both magnitude and phase images from a high-resolution, three-dimensional (3D) fully velocity-compensated gradient-echo (GRE) sequence. ^[6] A phase mask is created from the MR phase images, and multiplying these with the magnitude images increases the conspicuity of the smaller veins and other sources of susceptibility effects, which is depicted using minimal intensity projection (minIP). ^[7] Although the term SWI has been used by various authors to indicate sequences that are sensitive to T2FNx01 GRE techniques ^[8] or to high-resolution blood oxygen level-dependent venography, ^[5] our study has used SWI to refer to the use of magnitude or phase images, or a combination of both, obtained with a 3D, fully velocity-compensated, GRE sequence. Earlier studies have shown that the magnitude and phase information of SWI offers improved sensitivity, revealing low-flow vascular malformations that are invisible on conventional GRE sequences. ^[5],[9] The technique has been found useful in detecting developmental venous anomalies (DVAs) and cavernomas. ^[10],[11] There are also reports that show that high-flow lesions within the DVAs can be missed in SWI. ^[12] Essig et al. showed that high-resolution MR venography is superior to conventional TOF MRA in detecting AVMs and it is a useful technique in the treatment planning prior to surgery or radiosurgery. ^[13] We hypothesized that using the magnitude, phase, and minIP images of SWI we could understand the morphology and various components of the AVM better than conventional MRI sequences. Thus the aim of our study was to evaluate the imaging appearances of AVMs using SWI and to assess its potential in delineating the various components of AVMs.

Materials and Methods

Fourteen patients with brain AVMs (nine male and five female, mean age of 28.57 years; range 11-51 years) who had undergone MRI, including SWI, during the period January 2008 to December 2008 were included in the study. Institutional Review Board approval and informed consent were waived for this retrospective study. Clinically four patients presented with hemorrhage, four with headache, and three with focal neurologic deficit and seizures each. All the patients underwent computed tomography (CT) scan, MRI, and DSA. None of the patients had undergone any form of therapeutic intervention before imaging. Fourteen AVMs were found in 14 patients, of which eight were <3 cm and six were 3-6 cm in size. Spetzler-Martin grading was: Grade 1- one, Grade 2 - seven, Grade 3 - four, and Grade 4 - two.

Imaging

Imaging was performed using a 12-channel matrix coil on a 1.5T clinical scanner (Avanto, SQ Engine; Siemens, Erlangen, Germany). The sequences included: axial SWI, T1-, T2-, proton density (PD)-,and postcontrast T1-weighted images. All the sequences were acquired before the administration of contrast. The SWI sequence parameters were TR (repetition time), 48 ms; TE (echo time), 40 ms; Flip angle, 20°; bandwidth, 80 kHz; slice thickness, 2 mm, with 56 slices in a single slab; matrix size, 512 Χ 256. The sequence was fully flow compensated. A TE of 40 ms was chosen to avoid phase aliasing, and a flip angle of 20° was used to avoid nulling of the signal from pial veins located within the cerebrospinal fluid (CSF). ^[9] The acquisition time was 2.58 min with the use of iPAT factor-2.

The postprocessing involved three steps and has been described previously. ^[11] The first step was to remove incidental phase variations in the images due to static magnetic field inhomogeneity. This was done by means of a homodyne processing of the phase images, which uses a low spatial frequency phase map generated from the raw data itself. The Hamming-filtered reconstructed phase images were then subtracted from the original phase images using complex division. As a second step, the phase mask was created and unimodal signal processing was done. The phase mask was then multiplied with the magnitude data to enhance the visualization of vessels. Finally, 9 12-mm thick minIP images were generated every 2 mm. The entire image processing was automated and similar in all the patients. The phase, magnitude, and minIP images were available in the main console.

Data analysis

For each patient, the conventional MR images, including axial T1-, T2-, PD- and postcontrast T1-weighted images were analyzed first. This was followed by the evaluation of the SWI images, including phase, magnitude, and minIP. The analyses were carried out independently by two experienced radiologists. In case the findings differed, common consensus was reached by discussion. In each case, the features noted included localization of the AVM and delineation of its vascular components. The appearance of the various AVM components, including feeding artery, nidus, and draining vein, were noted using the conventional sequences and then using each of the SWI images, that is, phase, magnitude, and minIP. The presence of hemorrhage or calcification, if any, was also noted. Signal intensity of the structures was compared with that of the normal appearing white matter, that is, the structure with lower signal intensity than the normal appearing white matter was considered hypointense and vice versa. The structures that were homogenously hyper- or hypointense were designated hyper- or hypointense, respectively. The structures with areas of both hyper- and hypointensity were designated heterogenous. Visualization of these vascular structures on SWI was compared with that in conventional MR sequences. In all the patients, analyses of previously performed CT scans were also done to look for areas of hemorrhage or calcification. Pre- and postcontrast CT was performed on CT/I single-slice helical CT scanner (GE Medical Systems, Milwaukee, WI) with following parameters: 120 kV, 240 mA, 5-mm thick slices with 5-mm interslice spacing, 1 s rotation time, 20-cm field-of-view, and coverage from base of skull to vertex using intermittent axial slices. All the patients had also undergone a conventional angiogram within 1 month of MRI. The findings on MRI were confirmed on conventional angiogram.

Results

Analyses of the conventional sequences showed all the three components (feeding artery, nidus, and draining vein) of the AVM in all the patients except in three. In one patient the feeding artery could not be visualized, while in another two the draining veins were not identifiable. All the three components of the AVM appeared hypointense on PD-weighted sequence, while the adjacent gliosis appeared hyperintense. The hemorrhagic residue and calcification added heterogeneity to the appearance of nidus, but it could not be identified with confidence in most patients and was inseparable from the adjacent gliosis on the PD-weighted sequence. Analyses of additional sequences (T1- and T2-weighted imaging) and/or corroborative CT scan helped at identifying the gliosis and calcification. Finally, no possible hemodynamic information could be inferred from the conventional sequences.

Analyses of the SWI images showed the feeding artery in 13 of 14 patients in both magnitude and phase images and in ten of 14 patients in the minIP images. One patient in whom feeding artery was not seen in PD-weighted imaging, it was also not identified on any of the SWI images. On magnitude images the feeding artery appeared hypointense in eight, heterogenous in two, and hyperintense in three patients, whereas on phase images it appeared hyperintense in six and heterogenous in seven patients. In ten patients, in whom the feeding artery appeared hypointense or heterogenous on the magnitude image, the corresponding minIP images showed the feeding artery to be hypointense [Figure - 1]. In three patients in whom the feeding artery was hyperintense on the magnitude image, it could not be visualized on the minIP images [Figure - 2]. In one patient the feeding artery was not visualized on any SWI images. Thus in four patients the feeding artery was not visualized on minIP images, including one not visualized on any SWI sequence. In the other three patients, in whom the feeding artery was not seen on minIP, it had hyperintense appearance on both magnitude and phase. The nidus was identified in all the patients on magnitude image. It appeared hyperintense in eight patients [Figure - 3] and showed distinct hyperintense foci with adjacent or interspersed hypointensities in six patients. CT scans of these six patients revealed previous hemorrhage in five, and calcification in one patient [Figure - 4]. The nidus could be identified in the phase images in all the patients, though less distinctly than in the magnitude images, and it showed hyperintense (in nine patients) or heterogenous (in five patients) signal intensity. Nidus was identified in eight patients on the minIP images. This included two patients without and six patients with either hemorrhage or calcification. No distinction between the nidus, hemorrhage, or calcification was possible in the minIP images as all appeared hypointense in signal intensity.

The draining vein was seen in 12 patients in magnitude images, which appeared either hyperintense (in six patients, [Figure - 1]) or heterogenous (in six patients) in signal intensity. Similarly, it could be visualized in 12 patients on the phase images (hyperintense in four patients and heterogenous in eight patients), but less distinct compared with the magnitude images. In the remaining two patients, the draining vein was not visualized in any sequence including conventional as previously stated. In the minIP sequences, the draining vein was visible in only two patients, where it appeared hypointense in signal intensity. The results have been summarized in [Table - 1] and [Table - 2].

Discussion

AVM is the most common intracranial vascular malformation and is one of the common causes of spontaneous intracerebral hemorrhage in an adult. Hemorrhage is the commonest presentation followed by seizures, headache, and focal neurologic deficit. Mean age of presentation is 30-40 years with no sex predominance. There was male predominance in our study and it may probably related to the small sample size. Knowledge of the precise location and definition of the AVM including differentiation between its various components (nidus, feeding arteries, and draining veins) is important prior to planning the treatment. ^[13],[14] DSA is the gold standard for this, however, CT angiogram and MR angiogram can also give useful information. TOF MRA, contrast-enhanced T1-weighted images and high-resolution MR venography have been useful in AVM evaluation in the past. Each of these techniques has its own advantages and disadvantages in imaging the AVM. Disadvantage of TOF MRA include its dependence on the presence of vessels with high blood flow velocity and its inability to detect small AVMs. ^[13] Small AVMs could also be detected using contrast-enhanced T1-weighted images due to T1 shortening, independent of flow velocities. Its limitations include the use of contrast media, slice thickness, and a recurrent flow void phenomenon in high-flow vessels. ^[15] In the study by Essig et al., high-resolution MR venography failed to detect the feeding arteries in about half the patients. They therefore suggested its use in combination with the conventional MRA techniques, which emphasize the arterial component. In addition, high-resolution MR venography also failed to delineate exact size and shape of nidus as it suffered from the effects of previous hemorrhage. ^[13]

SWI is a new noninvasive imaging technique that uses tissue magnetic susceptibility differences to generate a unique contrast. Our study shows that in comparison with the conventional sequences, SWI showed no significant increase or decrease in the size of the nidus of AVM except in patients with either hemorrhage or calcification. In the presence of these factors, the prominent susceptibility effect was highlighted on the SWI, especially on minIP images. In the study by Essig et al., the minIP images analyzed showed the nidus with varying grades of conspicuity in all AVMs. In our study, the nidus was seen in only eight of the 14 patients. Of these, five patients had either hemorrhage or calcification, which probably was the factor responsible for the conspicuity and/or overestimation of the nidus size. Using the minIP, the feeding artery was seen in ten patients (71%) in our study compared with 47% in the previously reported study (13). The nonvisualization of AVM is probably due to the flow phenomenon from signal dephasing due to pulsatility and/or turbulence. The draining vein could be identified only in two (14%) compared with 80% in the previous study, (13) possibly owing to a wash-in of oxygenated blood into the venous structures seen in high-flow fistulas within the malformation. Overall therefore the minIP images did not prove as effective in identifying and differentiating the various components of AVMs in the majority of patients. It did, however detect hemorrhage/calcification effectively.

The images that we found most useful in separately identifying the various components of the AVM were the magnitude images. Irrespective of the possible hemodynamic variations in the 14 AVMs we analyzed, the AVMs and the components could be identified in all using magnitude except in one where feeding artery and two where draining vein could not be visualized. The nidus was seen predominantly hyperintense in all the patients in the magnitude image. This enabled differentiation from the hemorrhage and calcification seen as areas of hypointensity. The feeding artery was hypointense and the draining vein hyperintense on magnitude images. Thus the magnitude images gave the best visual perception of the various AVM components and helped identify hemorrhage and calcification. While the phase images detected the AVM in all the patients, the conspicuity and precise visual delineation of its components was less than optimum in most. The analysis of SWI images in AVM should thus include not only the minIP images but also the magnitude and phase images. The likelihood of picking up the various AVM components increases by analyzing all the 3 images.

The difference in signal intensities of the various components of the AVM in the SWI images can be due to several reasons. The field distribution inside a voxel that is traversed by a vessel produces dephasing of the spins. This loss of spin coherence is characteristic for the vessel orientation, blood oxygenation, the volume fraction occupied by the vessel, field strength, and voxel geometry. ^[16] The resonance frequency inside a blood vessel that is more paramagnetic than its surroundings depends on the orientation of the vessel with respect to the magnetic field. ^[17] Hence an AVM with vessels behaving as differently oriented channels of different sizes and blood oxygenation can have different signals based on the morphology.

One of the major limitations of this study is the small sample size. Further larger study with angiographic correlation is needed to understand the variation of signal characteristics of AVM in SWI. Since the low- and high-flow components behave differently in SWI, it may be appropriate to study the signal characteristics of such AVMs separately using SWI sequence. This in turn may give important hemodynamic information of the AVM noninvasively, which is needed for the management of these patients. We conclude that SWI sequence is an important tool in the workup of patients with AVM prior to surgery or radiosurgery. Magnitude images help in differentiating the different components of AVM and also helps in differentiating nidus from hemorrhage and calcification.

References

Copyright 2010 - Neurology India

The following images related to this document are available:

Photo images