Introduction

The small size of the structures in the spine [spinal cord 10–12 mm of diameter, spinal anterior artery 0.2–0.8 mm, Adamkievicz artery (AKA) 05–1.0 mm] [1, 2] makes it difficult to perform magnetic resonance angiographic (MRA) sequences in the spinal cord . The use of a higher magnetic field significantly improves the quality of images with a reduction in acquisition time [3]. This enables high resolution images of the different vascular phases (arterial, venous and late) to be obtained that are necessary for the study of the normal vessels and of vascular malformations in the spinal cord.

Materials and methods

We have examined 17 patients (for details see Table 1), eight female and nine male, aged 26–82 years, median age 52 years on a 3-Tesla MR system (Trio, Siemens, Forchheim, Germany); the patients were referred between January 2006 and October 2008 to the Department of Neuroradiology, Geneva University Hospital, with known spinal problems. The patients underwent the following imaging procedures after having been informed and according to our institutions guidelines. The protocol used was composed of sagittal T2-weighted imaging (T2WI) (TE 120, TR 3,500, 4-mm thickness, nine slices) and axial T2WI (TE 120, TR 4,000, 4-mm thickness, 20 slices), sagittal T1WI (TE 13, TR 626, 4-mm thickness, nine slices), diffusion tensor imaging (DTI) with 30 directions, three-dimensional (3D) MRA sequences (see details below) and fat-saturated T1WI in the sagittal (TE 12, TR 400, 4-mm thickness, nine slices) and axial (TE 12, TR 400, 4-mm thickness, 20 slices) planes after administration of contrast.

Table 1 Patient imaging findings (AKA Adamkievicz artery, AVM arteriovenous malformation)
Full size table

Three MRA acquisitions based on a 3D contrast enhanced gradient echo sequence were sequentially performed (Fig. 1) after a bolus injection of an intravascular contrast agent (0.2 mmol/kg, Vasovist; Bayer Schering, Berlin, Germany). The sequence was fat saturated and therefore it was not centrically reordered. A test bolus was then performed in order to determine the arrival time of the contrast agent in the arteries feeding the spinal cord, and the acquisition was started adequately so that the acquisition of the k-space centre matched the bolus arrival time. For the arterial and venous 3D gradient echo MR sequence, the parameters were TE/TR/flip angle = 1.52 ms/3.98 ms/20°, a matrix size of 384, pixel size = 1.2 × 1.0 × 0.9 mm, field of view (FOV) = 380 mm, no rectangular FOV, 88 slices, 0.9-mm slice thickness, bandwidth (BW) = 410 Hz/px. The phase encoding direction was chosen in the cranio-caudal direction to avoid flow and motion artefacts from the ascending hyperintense aorta projecting on the spinal cord, with an additional 33% oversampling. The sequence was acquired at the isocentre and required 1 min for a single phase.

Fig. 1
figure 1

A 29-year-old man with AVM: dynamic MRA, arterial-phase (a), venous-phase (b) and high-resolution images (c) with 3D reconstructions (d)

Full size image

For the high resolution late enhanced 3D gradient echo MR sequence, the parameters were TE/TR/flip angle = 1.89 ms/4.95 ms/20°, two averages, a matrix size of 512, pixel size = 0.7 × 0.6 × 0.6 mm, FOV = 330 mm, no rectangular FOV, 72 slices, 0.6 mm slice thickness, BW = 410 Hz/px. The sequence was also performed systematically at the isocentre. MPR reformatting and thin MIP reconstructions were realized for all acquisitions.

All cases, with the exception of AKA localization and patients with ischemic lesions, had a conventional angiography. The tumours were proven by histology.

Results

In all patients the main spinal vessels were seen and abnormal vascular lesions were visualized (see Table 1).

In two out of three patients examined for localization of the AKA, the visualization was easily done (Fig. 2), whereas in the third patient differentiation between artery and vein was more difficult, but possible by performing maximum intensity projection (MIP) and multiplanar reconstruction (MPR).

Fig. 2
figure 2

A 30-year-old woman: the MRA shows clearly an AKA (arrows, a) passing over the cyst (asterisk, a, b)

Full size image

In four cases of dural fistula the exact localization of the fistula was possible (Fig. 3).

Fig. 3
figure 3

A 57-year-old man with a dural fistula at the level of Th9 right illustrated by MRA (a) and angiography (b)

Full size image

In three cases with AVM, the level of malformation and the intra- and perimedullary localization was done as well as the arteries implicated (Fig. 4). The little shunts inside the nidus could not be individualized.

Fig. 4
figure 4

A 26-year-old woman with a cervical AVM: the MRA (a) showing the feeder, nicely demonstrated in the 3D reconstruction (b), corresponding to what is seen on DSA (c)

Full size image

Cavernomas are not seen on MRA and on conventional angiography; therefore, the realization of angiographic sequences is not necessary. In tumours such as hemangiomas or hemangioblastomas, hypervascularization was detected in addition to morphology and location of the masses.

In ischaemia [4], the parenchymal anomalies were visible early on diffusion imaging but unfortunately the implicated vessel was not detectable by MRA.

Discussion

The main advantage of spinal MRA is a better image quality and resolution, which improves the visualization of small-size vessels. The utilization of high vascular remanence contrast agent permits visualization of the arterial and venous phases as well as to obtain high-resolution late enhanced 3D gradient echo MR images.

Currently we could detect both the normal arteries (i.e. the AKA, Fig. 2) and the abnormal vessels as in the prior studies realized at 1.5 Tesla [5, 6]. While the “gold standard” for the evaluation of spinal vessels is conventional angiography, this procedure is still accompanied by risks and requires experience to perform. Dynamic MRA avoids an unnecessary angiography with superselective injections at numerous levels in the case, for example, where a dural fistula (Fig. 3) or an AVM (Fig. 4) are suspected [7–9]. Usually in our hospital, only the overlying and underlying segments are catheterized for the study of eventual feeder arteries. Krings [10] recommended the utilization of sequences such as CISS 3D or FIESTA, but the problem with these sequences is that differentiation between the veins and the arteries is not possible. When the localization of AKAs is necessary for the surgery of tumours or in the case of aortic aneurysm [11, 12] we avoid the risk of plaque embolization by not performing conventional angiography.

The real advantage of 3D dynamic vascular sequences is the capacity to non-invasively evaluate the shunt location which is extremely helpful to guide invasive conventional angiography [2]. Bowen and Kochan [13] described in 1995 that spinal contrast-enhanced MRA has greatly contributed to localizing these lesions and helping to avoid unnecessary multiple superselective injections; the drawback of the technique used by this author is the non-utilization of the dynamic phase component essential for the classification and subsequent treatment of the vascular malformations.

Recent technical improvements in contrast-enhanced MRA ,such as the use of SENSE or other parallel imaging techniques at 3 T, have allowed to greatly accelerate imaging, thus obtaining both a high spatial and temporal resolution [7, 9]; this can help to better depict feeding arteries as well as anatomy and flow velocity of early venous drainage, thus allowing improved classification of vascular malformations. Also, due to the high resolution of the technique, in the case with a very small structure such as a spinal cord cavernoma, the combination of MR and MRA allowed to help exclude another kind of vascular malformation.

Finally MIP and MPR reconstructions are crucial for the detection and analysis of the vascular malformation; for this purpose it is necessary for the physician to be trained in this kind of reconstruction.

Conclusion

Currently, MRA can provide the exact localization of a vascular malformation and AKA, and thus help detect and characterize spinal vascular lesions before conventional angiography for the diagnosis, thus avoiding an unnecessary intervention and decreasing the risk of embolism or dissections. Standard angiography should be used only for therapeutic purposes.