New MR sequences in daily practice: susceptibility weighted imaging. A pictorial essay

Sporadic cerebral amyloid angiopathy (CAA) is a common small-vessel disease associated with ageing, dementia and Alzheimer’s disease, that can only be diagnosed histopathologically following biopsy or at post-mortem examination. It consists of deposition of amyloid protein within the small and medium-sized cerebral arteries, which is likely responsible for increased vessel fragility and it is one of the major causes of lobar intraparenchymal haemorrhages in the elderly [8].

Computed tomography (CT) and conventional MR techniques are usually unable to show cerebral microbleeds (CMBs), which can be frequently observed on T2*-weighted gradient-echo MRI and have a typical lobar distribution [9]. Recent findings indicate that CMBs in the general elderly population are relatively common and are even more frequently observed in patients with AD [10].

SWI is much more sensitive than GE sequences in demonstrating CMBs in and around the arteriole vessel wall, which appear as black dots better displayed by minIP images (Fig. 2). Histologically, they represent perivascular clusters of haemosiderin-laden macrophages resulting from leakage from cerebral small vessels [11]. Higher field strengths, as expected, can improve CMB detectability, increasing the lesion’s contrast enhancement [7].

CAA is also characterised by white matter hyperintensities on conventional MRI sequences, which have been associated with cognitive impairment [12]. Vascular amyloid deposition is believed to be involved in the pathophysiological mechanisms that determine white matter hypoperfusion through either vessel stenosis or vascular dysfunction [13].

CAA should also be suspected in elderly patients with clinical signs of a progressive encephalopathy syndrome with seizures, in which extensive white matter abnormalities are discovered in brain MRI together with multiple CMBs with lobar and subcortical distribution (Fig. 2). It has been recently reported that this combination of MR findings should be interpreted as CAA-related inflammation, which can be treated with steroid therapy with a prompt resolution of the symptoms [14]. In those cases, the demonstration of an APoE ɛ4ɛ4 genotype can definitely support the neuroradiological diagnosis. Hypertensive encephalopathy and especially posterior reversible encephalopathy syndrome (PRES), when the signal abnormalities of the white matter are more symmetrical and located in parieto-occipital regions, should be taken into consideration for differential diagnosis.

Hypertensive encephalopathy is characterised by multiple CMBs which are usually silent and can be discovered when the patient is investigated with MRI in order to understand the cause of an intraparenchymal haemorrhage located outside the basal ganglia. SWI is more sensitive to CMBs than T2*-weighted GE sequences in blood pressure (BP)-related small vessels disease. CMBs are usually discovered both in deep basal ganglia and subcortical white matter (Fig. 3). Histologically they represent focal accumulations of haemosiderin-containing macrophages in the perivascular space of small brain vessels, indicating previous extravasation of blood, and are often associated with the presence of a symptomatic haemorrhage in the corresponding area. The number of CMBs, which remain detectable for years, is significantly associated with BP levels. Moreover, the presence of deep CMBs can be a useful marker of BP-related small vessels disease, helping in the differential diagnosis with CAA[15].

Iron deposition increases in the brain as a function of age, primarily in the form of ferritin and particularly in oligodendrocytes, but also in neurons and microglia. Typical sites of iron deposition include the globus pallidum, substantia nigra, and red and dentate nuclei. Ferritin is paramagnetic and produces strong susceptibility effects on T2*-weighted images. SWI filtered-phase images are particularly suitable for showing increased iron content in the brain ([16]. Abnormally elevated iron levels are evident in many neurodegenerative disorders, including Parkinson’s disease, Alzheimer’s disease, Huntington’s disease and amyotrophic lateral sclerosis.

The ability to measure the amount of ferritin in the brain can be used for a better understanding of the progression of the disease and is also helpful in predicting the treatment outcome.

Phase images allow a better distinction between the pars compacta and the pars reticulata of the substantia nigra, which contains iron (Fig. 1). As in patients with idiopathic Parkinson’s disease, there is evidence of increased iron in the substantia nigra [17], SWI sequences can be proposed as a useful imaging tool to identify iron deposition as a biomarker for disease progression, although longitudinal studies are required to support the usefulness of this specific application [18]. Accurate localisation of the subthalamic nucleus can also be achieved in the SWI phase maps at 3 T, allowing safe direct targeting for placement of electrodes in the treatment of Parkinson’s disease [19].

In 70–98% of patients with Alzheimer’s disease, intravascular amyloid deposition is found at autopsy [20] and CMBs are commonly observed. Recent evidence has shown that patients with multiple CMBs have more white matter hyperintensities and perform worse on mini mental state examination compared with patients with no CMBs, despite similar disease duration [21]. A significant correlation has also been reported recently between patients with at least one hypointensity in GRE-T2* imaging and those homozygous for the apolipoprotein E ɛ4ɛ4 gene, a well-known risk factor for Alzheimer’s disease (Fig 4) [22].

SWI has been demonstrated to be very useful in the acute phase of stroke for several reasons. First of all, it is very sensitive to the presence of CMBs, whose early identification is believed to predict the probability of potential haemorrhagic transformation after thrombolytic treatment [23]. It has also been hypothesised that CMBs can represent a link between cerebral haemorrhage and ischaemia [24].

SWI is also capable of identifying the acute intravascular clot in the main and distal branches of the cerebral arteries [25].

The evaluation of the intravascular content of deoxyhaemoglobin in hypoperfused brain is one of the most interesting applications of SWI. This technique has been proved to be useful for the assessment of tissue viability in patients with hyperacute cerebral ischaemia, as it provides functional information about the ischaemic penumbra. In acute arterial stroke, flow reduction or absence causes an increase in oxygen extraction, leading to a growth in the amount of deoxyhaemoglobin in the hypoperfused brain tissue. Deoxyhaemoglobin represents a supplementary source of contrast enhancement in SWI because of its paramagnetic properties, which determine a reduction in T2* as well as a phase difference between the vessel and its surrounding parenchyma. It appears markedly hypointense on both magnitude image and phase images. As the T1 and T2 properties of blood depend on the oxygen saturation of the blood, there are differences between arterial and venous vessels that cannot be discovered with conventional sequences. The visualisation of draining veins within areas of impaired perfusion allows the identification of penumbral brain tissue in a different way from the current perfusion-weighted imaging (PWI) techniques (Fig. 5) [26].

SWI has become a functional method for evaluating cerebral venous sinus thrombosis (CVST) by demonstrating engorgement of the venous system as a result of venous hypertension and collateral slow flow. Dural sinus thrombosis causes an increase in deoxyhaemoglobin concentration in the veins involved that appears as a prominent area of hypointense signal intensity on SWI; if CVST is treated successfully, this effect will disappear [26]. SWI is also capable of identifying the thrombosed pial vein, which is responsible for the development of oedema into the white matter, due to its high sensitivity in the demonstration of an intravascular clot. SWI can depict at the same time the features of parenchymal or extra-axial haemorrhages that can occur in the case of infarction, with higher sensitivity compared with conventional GE sequences (Fig. 6).

Cerebral arteriovenous malformations (AVMs) are easily displayed by conventional MRI and MR angiography because of their characteristic high flow, whereas venous malformations such as cavernomas, developmental venous anomalies and capillary telangiectasias cannot be adequately visualised without contrast medium administration and GE sequences, as they mainly consist of slow-flow small vessels. MR angiography techniques are often inadequate for visualising small vessels with slow flow. On the contrary, SWI sequences are well suited for the visualisation of very small vessels, such as the caput medusae of venous angiomas and telangiectasias as a result of a combination of slow flow with changes in deoxyhaemoglobin concentration [5]. The combined information derived from both phase and magnitude images is responsible for enhanced visualisation of such lesions.

SWI plays a substantial role in the identification and characterisation of cerebral vascular malformations, first of all improving their detection rate. On the other hand, allowing simultaneous visualisation of the different compartments of cerebral AVMs and of the relationship with one another and with the brain parenchyma, it is particularly valuable for therapeutic planning. The clinical usefulness of a similar technique, BOLD MR venography, was first demonstrated by Essig et al. [27].

Phase changes caused by the susceptibility differences between oxygenated and deoxygenated red blood cells and the different relaxation rates between venous and arterial blood provide a natural separation of arteries and veins. The 3D dataset of SWI can therefore be used as an alternative to MR angiography in order to better differentiate afferent arteries from draining veins. This advantage becomes relevant with small AVMs with recent bleeding, in which it is mandatory to identify the nidus in relation to eloquent areas in order to plan the best therapeutic approach, endovascular or surgical or both (Fig. 7). Veins appear dark due to T2* loss and processing of the phase image, while arteries show bright signal from time-of-flight (TOF) inflow enhancement. The use of higher flip angles and shorter TR make it possible to increase the contrast enhancement in the arteries without overly degrading the venography. For the evaluation of high-flow vascular malformations (i.e. AVMs and dural arteriovenous fistulas) the use of both minIP and MIP post-processing techniques can improve the image quality. The identification of small niduses, the exact location of the fistulous point and the depiction of the venous drainage patterns are the most relevant information that should be obtained from this technique. The elevated diagnostic accuracy of SWI compared with digital subtraction angiography (DSA) has been recently reported for the detection of arteriovenous shunting (AVS) in brain AVMs [28].

Dural AV fistulas (DAVFs) are usually more difficult to identify than AVMs on conventional MRI, especially when venous ectasia is absent, as in type I and II DAVFs; the patients have vague symptoms (i.e. headache or tinnitus) and there are no localising signs [29]. SWI is capable of better identifying the extent of the abnormal venous drainage compared with MR angiography. The conspicuity of the venous structures on SWI can be explained by the combination of a prolonged cerebral circulation time that favours increased oxygen extraction resulting in increased concentration of deoxyhaemoglobin, with the typical venous engorgement (Fig. 8) [30].

Concerning slow-flow vascular malformations (i.e. cavernomas, developmental venous anomalies and capillary telangiectasias) the phase information contained in SWI is responsible for an artefactual enhancement that dramatically improves MRI sensitivity to these pathological conditions. Without SWI, small vascular malformations could be entirely missed by conventional imaging techniques [31].

Individuals with cavernous malformations can present with epilepsy and focal neurological deficits or acute intracranial haemorrhage, although these vascular malformations are often discovered as incidental findings. Cavernomas that have previously bled are usually detectable on routine MRI because of the prominent signal intensity of haemorrhagic products. Cavernomas may also subsequently develop dystrophic calcifications that can be detected on CT. However, these lesions tend to be barely visible when they have not bled, except for faint enhancement after contrast medium administration that is often non-specific. Patients with multiple lesions can be diagnosed with familial cavernous angiomatosis and should be referred for genetic evaluation and counselling (Fig. 9). Individuals with symptomatic, growing, or haemorrhagic malformations should be considered for surgical resection. Close follow-up after diagnosis and treatment is helpful to identify lesion progression or recurrence [32].

Developmental venous anomalies (DVA) consist of a radially arranged venous complex converging on a centrally located venous trunk, which drains the normal brain parenchyma. They are better displayed with SWI than with T1-weighted contrast-enhanced images (Fig. 10).

Capillary telangiectasias are asymptomatic venous vascular malformations, which are smaller and less common than cavernomas and can sometimes occur in mixed cavernoma/telangiectasia lesions. These may occur sporadically or may be infrequently associated with syndromes, like hereditary haemorrhagic telangiectasia or ataxia telangiectasia. They may also manifest as a result of endothelial injury, such as radiation-induced vascular injury, particularly in children who have received cranial irradiation.

They are characterised by poor contrast enhancement and can therefore be missed on conventional T1-weighted and T2-weighted sequences (Fig. 11).

Ataxia telangiectasia is a rare autosomal recessive disorder characterised by an early-onset of progressive cerebellar ataxia, immunodeficiency, ocular and cutaneous telangiectasia and is associated with an elevated risk of intracranial tumours. SWI can discover brain gliovascular nodules with perivascular haemorrhage and small haemosiderin deposits presumably only in older patients, during the later stages of the disease; however, in our experience this technique can also be a useful screening tool in young patients (Fig. 12) [33].

MRI with GE sequences has been extensively used in the past for investigating young patients with traumatic brain injuries (TBI), either in the acute phase, when the clinical picture of a severe coma is not explained by CT, or months after trauma, in order to understand the causes of an unsuccessful recovery. In both cases, diffuse axonal injuries (DAI) can be responsible for the clinical picture. SWI has been widely used in the identification of smaller haemorrhages due to a shear-strain mechanism of injury, thus refining the prediction of outcome [34]. Haemorrhagic DAI lesions on SWI can be six-times better detected than T2*-weighted conventional GE sequences and the recognisable volume of haemorrhage is approximately twofold greater (Fig. 13). Reformatted SWI images in the sagittal plane are particularly helpful in the assessment of corpus callosum haemorrhages. Both number and volume of haemorrhagic lesions correlate with neuropsychological deficits [35].

Brain MRI predictors of tumour grade include contrast enhancement, oedema, mass effect, cyst formation or necrosis, haemorrhage, metabolic activity and cerebral blood volume. It is well known that the growth of solid tumours, such as gliomas, is dependent on the angiogenesis of pathological vessels. SWI can provide a thorough assessment of the internal angioarchitecture of brain tumours (increased microvascularity inside and beyond the tumour margins), together with the identification of foci of haemorrhage and calcification, thus representing an additional tool in the neuroradiological grading of cerebral neoplasms. The administration of a contrast agent (CE-SWI) allows discrimination among those three entities, as only blood vessels will change their signal intensity, while calcifications and regions of inactive haemorrhage (which can be differentiated from each other by the evaluation of phase images, as described above) will not. The clinical potential of contrast-enhanced BOLD MR venography at 3 T and 1.5 T for the study of brain tumours was first reported by Barth et al. [36], who demonstrated variable venous patterns in various types of tumours and in different parts of the lesions (oedema, contrast-enhancing areas, necrosis), which might represent increased blood supply and particular vascular patterns around fast-growing malignant tumours.

Kim et al. [37] recently assessed the added value provided by SWI in the differential diagnosis of solitary enhancing brain lesions compared with the use of conventional MRI alone. CE-SWI has been found to be equivalent or even superior to CE-T1 images in the evaluation of most tumours with necrotic areas (Figs. 14 and 15): the particular contrast combination within SWI images permits the simultaneous visualisation of the information otherwise obtained by a multimodal imaging approach including CT, CE-T1 SE, FLAIR and T2* conventional GE sequences [38].

Pinker et al. [6] demonstrated a correlation between intratumoural susceptibility effects, positron emission tomography (PET) results and histopathological grading. SWI has been proposed in the evaluation of clinical response to anti-angiogenetic drugs and in the differential diagnosis with pseudo-progression after chemo- and radiotherapy [39]. A correlation with MR PWI has also been attempted [26]. However, larger comparative studies of PWI and SWI are still needed to determine a more precise role of the new techniques in the grading of cerebral neoplasms.