Clinical 7 T MRI: Are we there yet? A review about magnetic resonance imaging at ultra-high field

Abstract

In recent years, ultra-high field MRI (7 T and above) has received more interest for clinical imaging. Indeed, a number of studies have shown the benefits from the application of this powerful tool not only for research purposes, but also in realms of improved diagnostics and patient management. The increased signal-to-noise ratio and higher spatial resolution compared with conventional and high-field clinical scanners allow imaging of small anatomical detail and subtle pathological findings. Furthermore, greater spectral resolution achieved at ultra-high field allows the resolution of metabolites for MR spectroscopic imaging. All these advantages have a significant impact on many neurological diseases, including multiple sclerosis, cerebrovascular disease, brain tumors, epilepsy and neurodegenerative diseases, in part because the pathology can be subtle and lesions small in these diseases, therefore having higher signal and resolution will help lesion detection. In this review, we discuss the main clinical neurological applications and some technical challenges which remain with ultra-high field MRI.

Introduction

MRI is a non-invasive medical imaging technique that provides both structural and functional data on human brain. In addition to being a non-invasive procedure, the degree of anatomical detail imaged with MRI makes this technique the modality of choice for the diagnosis, treatment-planning, and follow-up in a number of neurological diseases.

Since the initial application of MRI in medical diagnostics in the mid-1980’s, considerable efforts have been made by the scientific community in order to develop MR scanners with increasingly higher magnetic field strength (B₀) leading to enhanced signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and exquisite spatial resolution. These advancements allowed the gradual transition from the first grainy images obtained with 0.3–0.6 T MR scanner to the current extensive use in clinical practice of conventional (1–1.5 T) and high-field (3 T) MRI, whose main benefits include not only improved imaging quality and diagnostic accuracy, but also faster acquisition.¹

The first ultra-high field (UHF, or 7 T and greater) human MR image was acquired with the 8 T magnet at the Ohio State University in 1998.² The outcomes were exceptional and promising, resulting in the installations of numerous UHF research MR scanners worldwide, and in more recent years, the approval of clinical 7 T MRI scanners by the U.S. Food and Drug Administration (FDA). To date more than 80 human UHF MRI systems are operative, most of them with a magnetic field strength of 7 T.^3,4 Furthermore, after the FDA declared a non-significant risk for MRI up to 8 T in 2014,⁵ an important step forward for UHF MRI in October 2017 was the Magnetom Terra (Magnetom Siemens Healthineers, Erlangen, Germany) becoming the first 7 T MRI system to obtain FDA 510(k) clearance for clinical use in the United States, limited to the examination of the head, upper and lower extremities.⁶ This introduction will have remarkable repercussions in diagnostic radiology. In fact, the possibility to improve the spatial resolution reducing the size of the voxels thanks to the increased SNR of UHF, will enable clinicians to visualize smaller anatomical structures and greater detail of normal and pathological findings.

However, despite advances in magnet, coil, hardware and software technology, there are still limitations and technical challenges which need to be overcome.

In this article, we review the current literature and discuss the main clinical applications of UHF MRI in neuroimaging. Furthermore, we summarize the technical advantages and issues related with UHF MRI. A series of illustrative 7 T MRI examples is included.

References for this review were identified through searches of PubMed with the search terms “Utra-high field”, “7T MRI”, “7 Tesla MRI”, “diffusion MRI”, “Magnetic Resonance Spectroscopy”, “BOLD fMRI” cross-referenced with the terms “Multiple Sclerosis”, “cerebrovascular diseases”, “stroke”, “vessel imaging”, “brain tumor”, “epilepsy”, “neurodegenerative diseases”, “dementia”, “Alzheimer”, “Parkinson”. Only articles published in English were reviewed. Reference lists of identified articles, book chapters, and authors’ own references were also explored. Selection criteria were the novelty and importance in terms of potential clinical application of the reported results. Peer-reviewed articles published from 1967 until March 2018 have been included, in addition to two non-English historical references but with available translations dating back to the mid-1800s.

Clinical applications

There are a number of neurological diseases where UHF MRI has shown some benefit. These include multiple sclerosis (MS), cerebrovascular disease, neuro-oncology, epilepsy and neurodegenerative diseases. This may be because of the prevalence of these clinical entities, but also by the need to improve their diagnosis and clinical management through the identification and recognition of earlier disease and new imaging biomarkers. Furthermore, UHF MRI may play a role in helping neuroscientists, neurologists, neurosurgeons and neuroradiologists further insights on the pathophysiological mechanisms behind these conditions.

Multiple sclerosis

MS is the most common immune-mediated inflammatory demyelinating disorder affecting the central nervous system (CNS) and the leading cause of non-traumatic neurological disability in young adults in Northern America and Europe.⁷ The diagnosis of MS is primarily clinical, but sometimes it may be challenging and further examinations are required, including radiological and laboratory tests. MRI is a helpful tool to support the diagnostic process, revealing the radiological dissemination in space and time of MS lesions; moreover, it may provide information relevant not only to clinical follow-up and monitoring of response to treatment, but also to a better understanding of MS pathophysiology.⁸

MS lesions or plaques typically appear as focal areas of hyperintensity on dual-echo (proton-density and T₂ weighted) and fluid-attenuated inversion-recovery (FLAIR) imaging on MRI, while post-gadolinium T₁ weighted images allow active lesions to be distinguished from inactive lesions, the first being enhanced on MRI unlike the latter. Even if 7 T MRI has not yet resulted in an earlier diagnosis, recent studies have demonstrated the main advantages of imaging at ultra-high field in MS. For instance, one of the first studies comparing 3 T and 7 T MRI in MS showed that on 7 T magnetization-prepared rapid acquisition gradient-echo (MPRAGE) a significantly higher number of focal MS lesions was detected in areas defined as “normal-appearing white matter” (NAWM) using standard clinical 3 T FLAIR: these abnormalities may contribute to the pathological changes that have been reported in NAWM.⁹ Currently, it seems that there is increased sensitivity in the detection of gray matter lesions with UHF MRI^10,11 compared with 3 T MRI.

Another important benefit refers to cortical gray matter lesions’ detection. According to the 2017 revision of the McDonald criteria for the diagnosis of MS, cortical lesions can be used to fulfill MRI criteria for dissemination in space.¹² Currently, this type of lesion is not easily identified with standard MRI protocols, due to the smaller volume of gray matter and lower levels of inflammation, but in the past years UHF MRI has demonstrated to provide an improved characterization of cortical lesions thanks to the increased spatial resolution and enhanced contrast.^13–16 A recent post-mortem verification study showed that the use of a multicontrast protocol including T₁ weighted, T₂ weighted, FLAIR, double-inversion recovery (DIR) and T2* at 7 T provided a cortical lesions detection rate more than two times higher as compared with 3 T; however, a considerable part of pathologically proven cortical lesions still remained undetected at 7 T MRI.¹¹

Furthermore, UHF MRI has been increasingly used to investigate new possible hallmarks to accurately differentiate MS from its mimics. For example, various research groups have described central venules in MS lesions as a key element for the diagnosis of MS in challenging cases; this phenomenon, also named the “central vein sign” (CVS), was first reported by pathological studies in the 19th century,^17,18 and eventually confirmed using UHF MRI.^19,20 Even if the perivenular distribution of MS lesions is still detectable at T₂* weighted imaging with conventional and high-field scanners and high detection rate can be gained with optimized T₂* protocols,²¹ the enhanced susceptibility effects around blood vessels achievable with UHF MRI result in reduced echo times and high SNR, which in turn can be used to generate more detailed imaging of the relationship between veins and MS lesions.²² Indeed, Tallantyre et al demonstrated that the T₂* weighted imaging at 7 T has the highest sensitivity for central vein detection compared with 3 T,²² and it can be helpful to distinguish between patients with demyelinating MS lesions and incidental asymptomatic white matter lesions in subjects without MS.²³ Moreover, several 3 and 7 T MRI studies showed that the proportion of the CVS in patients with MS is significantly higher compared with white matter lesions in patients with other neurological diseases, including neuromyelitis optica spectrum disorder, systemic autoimmune diseases, cerebral small vessel disease, Susac syndrome, and migraine.²⁴ These are typical differential diagnoses for MS and can sometimes be difficult to differentiate clinically and on MRI.

Another novel radiological MS feature is the peripheral paramagnetic rim revealed by MR susceptibility imaging (T₂* weighted magnitude and susceptibility-weighted phase images) in a subset of chronic lesions. Whether this finding is caused only by iron deposition or other MS pathological processes is still debated, but in recent years relevant insights to its pathophysiological significance have been provided with UHF MRI.^20,25–27 Indeed, in addition to higher SNR and space resolution, phase images at UHF deliver contrast specific to field perturbations and show excellent contrast of local iron in white matter plaques, as susceptibility effects scale with field strength²⁵ (Figure 1).

Figure 1. .

Axial SWI minIP Images at 7 T (left) and 3 T (right) of a healthy volunteer. The depiction of vessels is noticeably superior at UHF. The higher susceptibility and spatial resolution at 7 T allows detection of smaller vessels and the deoxyhenoglobin in veins which may be important for the detection of the “CVS” in the differential diagnoses of multiple sclerosis compare to other diseases which can cause demyelination such as neuromyelitis optica spectrum disorder, systemic autoimmune diseases, cerebral small vessel disease, Susac syndrome, and migraine, which typically woud not have the CVS. Resolution 7 T: 0.2 x 0.2 x 1.5 mm³, scanning time = 5 min. Resolution 3 T: 0.9 x 0.9 x 1.2 mm³, scanning time = 5 min. CVS, central vein sign; SWI, susceptibility-weighted imaging.

Finally, a promising application of UHF MRI is MR Spectroscopy Imaging (MRSI). Indeed, the increased SNR results not only in improved spatial resolution for 7 T MRSI and shorter scan times, but also in enhanced spectral resolution (or chemical shift) compared with conventional and high-field 3 T clinical MRSI. The improved sensitivity and specificity of 7 T MRSI allow identification of metabolites with low concentration and offer improved discrimination of peaks of metabolites that otherwise overlap at lower field strength, such as glutathione (GSH), glutamate (Glu), glutamine (Gln), and myo-inositol (mI)²⁸ (Figure 2).

Figure 2. .

7 T MR spectroscopy with metabolite peaks. The single voxel spectrum (2 cm isotropic) is located in the posterior cingulate cortex. The green box on top demonstrates the residual of the LCmodel fit²⁹ .Reliable concentrations have been demonstrated, with a standard deviation (%SD) inferior to 20% for most metabolites (Table). The improved sensitivity and specificity of 7 T MRS allow identification of metabolites with low concentration and discrimination of peaks of metabolites that overlap at lower field strength, such as glutamate (Glu), glutamine (Gln), and myo-inositol (mI). SD, standard deviation.

GSH is an indicator of oxidative status in the human brain and a spectral editing scheme called band selective inversion with gradient dephasing using proton MRSI has shown a significant reduction of GSH between the gray matter in MS patients and normal controls, indicating the potential of GSH as a marker for disease phenotype in MS.³⁰

However, most of the results concerning these new potential MS hallmarks derive from small-cohort or single-centre studies; therefore large, prospective, multicenter trials are required to confirm their diagnostic role in MS.

In summary, the main advantages of UHF MRI for MS are the following: a better identification and morphological characterization of both white matter and gray matter lesions; the improved sensitivity to detection of new possible MS marker, including the CVS and the peripheral paramagnetic rim; the enhanced quantification of metabolites on MRSI. Although the role of these findings in MS diagnosis and clinical management still needs to be established, and the high cost of UHF MRI scanners makes unlikely this imaging technique to be routinely and widely used in clinical practice in the near future, we expect a growing application of UHF MRI in MS, not only for the validation and identification of new MS hallmarks in large clinical studies, but also to guide the development of novel diagnostic and therapeutic strategies at lower field.

Cerebrovascular disease

Worldwide, stroke is a leading cause of mortality and disability.³¹ Both CT and MRI have a critical role in the diagnosis, treatment-planning, and follow-up in stroke patients: although CT has wide availability and faster acquisition time, MRI is the best technique to accurately visualize and characterize acute stroke on diffusion as well as both large and small vessels in the brain.³² 3D time-of-flight (TOF)-MR angiography (MRA) is an established technique for imaging intracranial arteries thanks to the high signal contrast between the moving vascular protons and stationary protons. The T₁ relaxation times of tissues increase with field strength, so at UHF MRI, the longer T₁ values for tissue augment suppression of static background signal in TOF MRA increasing SNR and CNR in vessels with a resolution between 0.2 and 0.3 mm³.³³ (Figure 3). This allows detection of intracranial vessels with diameters less than 0.3 mm, including the small lenticulo-striate arteries, the pontine arteries, and the paramedian thalamic–subthalamic arteries, without ionizing radiation of invasive methods (digital subtraction angiography and CT angiography) or the intravenous administration of paramagnetic contrast agents such as gadolinium³⁴ (Figure 3).

Figure 3. .

Axial TOF-MRA MIP images at 7 T (left) and 3 T (right) of a healthy volunteer. An increased number of vessels is visible at UHF. Please note the increase conspicuity of the lenticulostriate arteries (arrows) arising from the M1 segement of the MCAs and the higher contrast seen in the insular branches of the MCAs (arrowheads). Resolution 7 T: 0.3 × 0.3 × 0.3 mm³, 4 × 72 slices, scanning time = 4 × 2.23 min. MCAs, middle cerebral arteries; MRA, MR Angiography; MIP, maximum intensity projection; TOF,time of flight.

Recently, a novel high-resolution black blood MRI technique has emerged for intracranial vessel wall imaging at 3 and 7 T using 3D turbo spin-echo (TSE) sequences with variable flip angles.^35–38 This T₁ weighted technique (also known as SPACE, Sampling Perfection with Application optimized Contrast using different angle Evolutions) offers an isotropic 0.5 mm spatial resolution, adequate flow suppression due to the long echo train, and near whole-brain coverage in less than 10 min; moreover, the excellent contrast between intracranial arterial wall and cerebrospinal fluid (CSF) allows better delineation of the vessel wall boundaries in vessels surrounded by CSF and to identify intracranial arterial disease.^37,38 At 7 T the substantial increase in SNR has led to significantly better 3D T₁ weighted SPACE image quality compared with 3 T, demonstrating an important potential for improved diagnostic performance^35,38 (Figure 4).

Figure 4. .

Thin minimum intensity projection across 10 mm slices from a 22-years-old healthy female subject at 3 and 7 T. Improved delineation of LSAs (white arrows) and reduced blurring, especially in the more distal vessels (arrowheads), can be appreciated at 7 T compared to 3 T. The visualization of these LSAs allows for visualization of normal vessels as well as the detection of possible pathologies such as arterial dissection and small LSA aneurysms (so called Charcot Bouchard microaneurysms). LSA, Lenticulostriate artery.

MRA and T₁-SPACE at UHF may provide novel clinically relevant non-invasive approaches to improve assessment of cerebrovascular diseases, including intracranial aneurysms,^39–41 vessel stenosis,^42,43 and microvascular pathological changes as well.⁴⁴

Cerebral small vessel disease (SVD), a heterogeneous group of diseases affecting the perforating arterioles and capillaries in the brain, is the principal contributor to stroke and vascular cognitive impairment and dementia.^45,46 Features of SVD on MRI include brain atrophy, lacunes, enlarged perivascular space (PVS), white matter hyperintensities (WMH) and cerebral microbleeds. UHF has demonstrated potential clinical utility in better depicting these parenchymal lesions. T₁ weighted MPRAGE at 7 T offered superior visualization of the internal structure of stroke lesions and allowed improved detection of PVS compared with 3 T.⁴⁷ Also T₂ weighted TSE sequence at 7 T showed the same benefits, even though the authors observed more motion artifacts.⁴⁷ Comparable results in the identification of WMH were achieved with 3 and 7 T, yet better contrast between WMH and healthy tissue was observed at UHF.⁴⁷ Furthermore, a significantly higher number of cerebral microbleeds was identified at 7 T compared with both 3 and 1.5 T MRI.^48–50 Finally, the detection of cortical microinfarcts, another common SVD lesion not reliably visible on clinical MRI, was demonstrated to be feasible with 7 T scanners, providing new opportunities to investigate their role in cognitive function of SVD patients.⁵¹ While it’s more evident the relevance of these preliminary findings in research, further investigations are required to understand their real implications for routine clinical practice, as previous studies included a small number of patients.

Blood oxygenation level dependent functional MRI (BOLD fMRI) is another imaging technique showing significant advantages at UHF, derived from the increased SNR and susceptibility effects as well. The two main applications of BOLD fMRI at 7 T included the discrimination of the functional response between different cortical layers and mapping patterns of neuronal population activity for the first time even to submillimiter isotropic resolution.^52–55 Currently, the use of BOLD fMRI at UHF in clinical practice is still limited, but in the future it might be possible to use this tecnhique clinically, as for tracking cortical reorganization during rehabilitation in stroke patients.

In addition to the advances in UHF MRI systems, new image analysis and post-processing techniques have been established for segmentation and quantification of microvessels, allowing characterization of their morphologic details such as vessel length, tortuosity, caliber, and ultimately leading to identification of diseased microvessels.⁵⁶

Neuro-oncology

Neuro-oncology is another important field where UHF MRI has shown encouraging promise for improving the diagnosis, treatment, and follow-up of brain tumors. For instance, increased spatial resolution may help in the detection of microvasculature in angiogenesis as well as small parenchymal and leptomeningeal metastases.

Moreover, the identification of metabolites with low concentration in 7 T MRSI is of remarkable interest for evaluating patients with glioma, the most common primary malignant brain tumor in adults.⁵⁷

Ex vivo studies have demonstrated that mIn/total Choline (tCho) could be used to differentiate active tumor growth from proliferation of reactive astrocytes in response to treatment such as surgery or radiotherapy, and that mIn/tCho is able to distinguish the low grade gliomas upgraded to grade III from the recurrent grade IV lesions.^58,59 More recently, an in vivo study has verified differences in metabolite levels for regions of tumor vs normal brain, as well as between lesions.⁶⁰ Furthermore, Verma et al demonstrated the ability of two dimensional localized correlated MR spectroscopy at 7 T to detect 2-hydroxyglutarate as a biomarker for the in vivo determination of IDH mutation status in gliomas.⁶¹ Based on these encouraging results, MRSI at UHF may represent a valuable non-invasive tool to accurately assess tumor classification and burden definition, leading to improvements in treatment planning and ultimately in patient prognosis and quality of life. The 2016 WHO classification of brain tumors made the identificaton of IDH mutation vs wildtype critically important for the subtyping of glioblastoma multiforme and the outcome.

Susceptibility-weighted imaging (SWI) and T₂* at 7 T have been investigated to study the intratumoral microvascular structure in gliomas. Susceptibility contrast of para-magnetic substances, such as deoxyhemoglobin, is magnified at UHF and allows a superior depiction of veins and microhemorrhages.^62,63 This is particularly useful in brain tumors, where increased metabolism leads to high deoxyhemoglobin levels and a better visualization of microvasculature on SWI. The analysis of tumor neovascularization is helpful to describe the grade of brain tumors and offers information on patient prognosis as well.^64,65 One study, e.g. suggested that the fractal dimension of intratumoral SWI pattern at 7 T MRI (indicative of microangioarchitecture and microbleedings) may effectively differentiate histopathological glial brain tumor grades.⁶⁶ Grabner et al explored the application of 7 T SWI to analyze changes in gliomas vascularization under antiangiogenic therapy, and they found that this task is especially appropriate for UHF MRI.⁶⁷ Another research group provided evidence for an improved depiction of tumor microvascularity at 7 T compared with 1.5 T, and showed increased vascularization from low- to high-grade gliomas.⁶⁸ Furthermore, the authors reported that the ring-shaped enhancement after contrast administration in a high-grade glioma is comparable at 1.5 and 7 T, but the central necrosis is emphasised at 7 T,⁶⁸ and this detailed characterization may be helpful to differentiate it from other similar radiological findings, such as necrotic metastases and cerebral abscesses. These studies demonstrate UHF MRI capabilities to provide useful information for grading gliomas and for monitoring tumor therapies.

Promising results of 7 T MRI for pituitary gland imaging have been achieved as well. Cushing’s disease is the most common cause of adrenocorticotrophic hormone (ACTH)-dependent Cushing’s syndrome, and is characterized by ACTH-producing adenoma located in the pituitary gland.⁶⁹ MRI is the preferred imaging technique to study pituitary gland in Cushing’s Disease, but the majority of pituitary adenomas are microadenomas, and they result undetected in 36–63% of patients scanned with conventional clinical machines.^70,71 Therefore, further examinations, including invasive procedures such as inferior petrosal sinus sampling (IPSS), are often required to obtain a final diagnosis.⁷² De Rotte et al demonstrated that more lesions were detected at 7 T MRI than 1.5 T, and the characterization of the lesions was more accurate at 7 T as well.⁷³ However, the authors reported the magnetic susceptibility effect related to the presence of air in the sphenoid sinus as a potential disturbing artifact for pituitary gland imaging.⁷³ Recently, Law et al reported a case of pituitary microadenoma that was visible at 7 T, but not easily seen on standard 1.5 T and even 3 T imaging studies⁷⁴ ; moreover, the 7 T MRI findings correlated with the results of IPSS⁷⁴ (Figure 5 ).

Figure 5. .

Postcontrast T₁-weighted MR images at 7 T (left), 3 T (middle), and 1.5 T (right). 7 T imaging demonstrates what appears to be an 8-mm right-sided hypoenhancing pituitary microadenoma (white arrow), not visible at 3 and 1.5 T. From Law et al., JNS 2018⁷⁴ (https://doi.org/10.3171/2017.9.JNS171969), permission from Elsevier.

In the future, 7 T MRI may become a routine valuable diagnostic technique in patients with MRI negative Cushing’s disease, possibly preventing the need of IPSS and improving the surgical planning and outcomes as well.

For brain metastases, the most common malignancy affecting the brain,⁵⁷ a study published in 2010 showed that the detection rate of cerebral metastases of bronchial carcinoma was almost equivalent on 1.5 and 7 T, but 20% more intralesional microhemorrhages were identified on SWI at 7 T.⁷⁵

Finally, a recent study reported that in two patients with orbital choroidal melanoma uncertainty of optic nerve involvement at 3 T which was superiorly depicted at 7 T. This had relevant consequences for clinical management.⁷⁶

Epilepsy

Epilepsy affects approximately 50 million people worldwide, and up to 30% of patients have disabling seizures that are refractory to antiepileptic drugs.⁷⁷ Focal cortical dysplasias (FCD) and mesial temporal sclerosis (MTS) are the two most common causes of drug-resistant focal epilepsy syndromes, and in these cases resective surgery is the most effective treatment for patients to become seizure-free, improving their quality of life.⁷⁸ The success rate of epilepsy surgery is critically related to the ability to identify a structural lesion on MRI; indeed, in patients with an MRI-detectable lesion, seizure freedom after surgery is achieved in around 70% of cases, compared with 40% when MRI is negative.⁷⁹ Currently, a significant proportion (almost 25%) of surgical candidates demonstrates no relevant structural MRI abnormalities with conventional scanners, so the introduction of more higher resolution imaging techniques and diagnostic tools is needed, not only for a better surgical outcome, but also for facilitating the pre-surgical planning without invasive procedures, such as intracerebral electrode implantation.⁸⁰ De Ciantis et al reported structural abnormalities detected at 7 T MRI and not previously identified with 1.5 and/or 3 T MRI in 6 out of 21 patients (29%) with focal epilepsy.⁸¹ Specifically, two sequences showed the advantage of 7 T: (1) the 2D T₂* weighted dual-echo gradient-recalled echo (GRE) targeted for localization of the seizure onset zone, improving the evaluation of the different components and even layers within the cortex, and (2) the 3D magnetization-prepared (MP)-FLAIR sequence.⁸¹ In a previous study by the same group, 7 T MRI was found to reveal more anatomic detail compared with 3 T in a group of 10 patients with polymicrogiria, a malformation of the cerebral cortex characterized by an excessive number of abnormally small gyri resulting in a large spectrum of possible symptoms, including refractive epilepsy.^82,83 Thanks to higher resolution 7 T MRI, the detection of bilateral involvement was demonstrated in four patients who had been classified as having only unilateral polymicrogyria at 3 T⁸² ; moreover, 3D susceptibility-weighted angiography (SWAN) at 7 T revealed numerous dilated cortical veins not visible at 3 T, suggesting a role for vascular dysgenesis in the pathogenesis of polymicrogyria.⁸²

Another recent paper investigated the role of PVS in the brain as a new potential biomarker for the altered macrophage activity associated with seizure onset at 7 T MRI.⁸⁴ The authors used axial T₂ weighted TSE sequences to analyze and quantify PVS in 21 subjects with focal epilepsy and 17 healthy volunteers: PVS distribution was found to be significantly more asymmetric in epilepsy patients, and the region of maximum asymmetry was within the suspected seizure onset zone in 72% of cases.⁸⁴ Thanks to improved contrast and resolution, the depiction of PVS, also known as Virchow-Robin spaces, is significantly improved at UHF and greater details can be visualized compared with 3 T MRI.⁴⁷

Springer et al demonstrated a higher diagnostic confidence for MTS at 7 T compared with 3 T, primarily due to the higher spatial resolution in the coronal T₂ TSE sequence, even if the difference did not reach statistical significance.⁸⁵ In their work, they were able to identify which hippocampal subfields were affected, toward assignment of the histologic subtype according to the ILAE consensus classification of hippocampal sclerosis.⁸⁶ The 7 T coronal TSE sequence demonstrated the best conspicuity for the depiction of hippocampal area, as the susceptibility artifacts commonly seen in the skull base region did not significantly affect the imaging of hippocampus.⁸⁵ However, the coil-related signal decrease in the posterior fossa as well as the higher B₁ inhomogeneity affecting imaging at UHF are still a problem.⁸⁵

Finally, UHF MRI may potentially provide further insights into the study of patients with cryptogenic seizures, that traditionally represent around 30% of all epilepsies.⁸⁷ For instance, previous studies have shown that SWI and T₂*-weighted GRE sequences at 7 T were able to identify more cerebral cavernous malformations and possibly associated developmental venous anomalies compared with 3 and 1.5 T clinical scanners, respectively, and these findings may have important implications for diagnostic and therapeutic purposes as well.^88,89

Neurodegenerative diseases

Alzheimer’s Disease (AD) is the most common cause of cognitive decline in the elderly.⁹⁰ A dramatic rapid increase in population numbers of older people worldwide is expected in the next decades thanks to the progress made in medicine and social conditions.⁹¹ Consequently, the incidence and prevalence of AD and other dementias will increase, with huge costs to society and healthcare systems.⁹¹ Despite the efforts made by the scientific community to better understand the neuropathology in AD, the pathophysiological mechanisms of this devastating disorder are still not well-known and therapies that effectively reverse or slow AD progression are desperately needed.⁹²

Neuroimaging approaches with UHF MRI may offer novel means to investigate AD. For example, previous studies showed that the sensitivity of 7 T MRI in detecting changes in the hippocampal strata is superior at 7 T compared with 1.5 and 3 T.^93,94 The stratum radiatum and lacunosum-moleculare of the Cornu Ammonis’ field 1 (CA1) have been shown to be thinner in mild AD patients than in controls, without any changes in the stratum pyramidale of CA1(93). Furthermore, the use of new segmentation techniques at 7 T allows superior quantitation of hippocampal subfields volumes because of the increase CNR, SNR, and spatial resolution. Investigators have demonstrated a decrease in the volume of all subfields (except CA2) and the enthorinal cortex in AD patients.^95,96 More recently, Sepehrband et al were able to image the hippocampal subregions, including the stratum pyramidal of rostral CA3, the alveus, and even the endfolial pathway⁹⁷ (Figure 6).

Figure 6. .

Hippocampal subfield imaging with high resolution coronal T₂ weighted contrast at 3 and 7 T. 3 T image: in-plane resolution: 400 μm, slice thickness: 2 mm, sequence: BLADE⁹⁸ , no repetition, total acquisition time: 10 mins. 7 T image: in-plane resolution: 300 μm, slice thickness: 2 mm, sequence: TSE, 4 averages and 2 concatenations, total acquisition time: 11 mins. Note that Fimbria (white arrow), Alveus (arrowhead), and Stratum layers were resolved at 7 T. SE, turbo spin echo.

The investigation of T₂* hypointensities has taken advantage of UHF MRI as well. Van Rooden et al reported that 3D T₂* sequences at 7 T demonstrated abnormalities in human brain specimens with AD and cerebral amyloid angiopathy (CAA), including hypointense foci and/or diffuse grainy inhomogenities of the cortex⁹⁹ ; furthermore, iron deposits were co-located with amyloid deposits on histology.⁹⁹ According to the authors, these foci may represent a neuroimaging biomarker for in vivo detection of amyloid-β plaques, facilitating the diagnosis of AD and CAA.⁹⁹ However, another ex vivo study using 7 T MRI showed that the hypointense foci were primarily located within the subiculum of AD specimens, suggesting that they may be more related to activated iron-containing microglia rather than amyloid deposits.¹⁰⁰

Recently, the role played by PVS as a clearance system of cerebral metabolic products and its implication for neurodegeneration have been investigated. The UHF MRI enabled to study these structures with a quantitative approach and explore their possible influence in dementia. One post-mortem study showed a clear positive association between dilation of juxtacortical PVS in a given region revealed on 7 T MRI and higher amyloid-β deposition in the same area of five cerebral AD specimens.¹⁰¹ Furthermore, Cai et al used the T₂-SPACE sequence at 7 T to compute the total PVS volume in five AD patients compared with three healthy controls, and they reported that PVS density in centrum semiovale was significantly higher in AD patients.¹⁰²

Finally, new evidence suggests that subtle white matter changes begin in preclinical AD and can be measured by diffusion tensor imaging (DTI), an advanced diffusion-weighted MRI (DWI) modeling technique for assessing the microstructure of the brain.¹⁰³ Compared to 3 T, the higher spatial resolution and reduction of partial volume effects at 7 T enables better separation of fiber bundles in white matter and cortical regions as well^104,105 (Figures 7 and 8). A multipurpose multishell DWI sequence was reported to fit a clinically acceptable acquisition time frame of ~10 min.¹⁰⁹

Figure 7. .

Color fractional anisotropy (A, B, D, E) and DTI principal direction of diffusion (C, F) maps from the Human Connectome Project. 7 T (A, B, and C, 1.05 mm³ isotropic resolution) and 3 T (D, E, and F, 1.25 mm³ isotropic resolution) from the same subject. Red, green and blue indicate regions with diffusivities oriented primarily laterolateral, ventrodorsal, and rostrocaudal, respectively. Improved resolution allows identification of fiber tracks (red arrows and yellow arrows in B) which are faintly visible at 3 T. Note increased B1 + inhomogeneity resulting in signal loss in temporal lobe regions at 7 T (white arrows in A).

Figure 8. .

Fiber orientation glyphs (A, B) and tractography reconstructions (C, D) from the Human Connectome Project. 7 T (A, C: 1.05 mm³ isotropic resolution) and 3 T (B, D: 1.25 mm³ isotropic resolution) from the same subject. Fiber orientations were estimated using FSL BEDPOSTX¹⁰⁶ and visualized using the Quantitative Imaging Toolkit (QIT);¹⁰⁷ the 7 T data shows improved modeling of cortical fiber orientations (blue arrow). Tractography models of the superior longitudinal fasciculus I were created using deterministic streamline tractography in QIT^107,108; the 7 T data shows improved reconstruction of orbitofrontal connections of the pathway in both hemispheres (white arrow).

Parkinson’s Disease (PD) is the second most common neurodegenerative disorder after AD,¹¹⁰ and it is clinically characterized by motor symptoms, such as resting tremors and rigidity,¹¹¹ and pathologically by α-synuclein intracellular inclusions, resulting in Lewy bodies formation and loss of neurons in substantia nigra (SN) and elsewhere.¹¹² Several 7 T studies of PD have demonstrated the advantages of UHF MRI in comparison with conventional scanners imaging, reporting improvements in diagnostic accuracy and treatment planning. For instance, Cosottini et al showed that 3D multiecho SWI sequence at 7 T was able to precisely characterize the SN and its internal organization, leading to near-perfect discrimination of PD patients from age-matched healthy subjects.¹¹³ Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is an effective surgical therapy for patients with advanced PD,¹¹⁴ however in some cases this treatment is not able to provide relief of parkinsonian symptoms, and possible side effects may occur as well.¹¹⁵ One of the most important factors for a successful outcome is the accuracy of targeting: the electrode should be placed preferentially in the motor region of the STN in order to exclude or minimize the stimulation of the non-motor zone.¹¹⁵ Plantinga et al demonstrated that tractography-based technique at UHF MRI may facilitate the identification of the motor zone of the STN in individual patients, allowing for more optimized patient-specific surgical planning.^116,117 Although 7 T MRI can provide significantly improved visualization of STN, geometric distortion (or pixel shifts) is more pronounced with increasing field strength, representing a critical challenge for DBS.¹¹⁸ Nonetheless, different approaches, including distortion correction and image reconstruction techniques such as QSM, have recently shown encouraging results and might be useful to address this issue.^119,120

Current challenges

Current major challenges in UHF MRI include inhomogeneity in the main magnetic field (B₀) and the applied radiofrequency (RF) field (B₁), increased specific absorption rate (SAR), and the increased sensitivity to motion artifacts.

B₀ inhomogeneities are related to the magnetic susceptibility difference between air cavities in the human skull and brain tissue. This effect scales linearly with the magnetic field strength, leading to significant distortion and non-uniformity of signal intensity in images obtained at UHF, especially when very fast image acquisition techniques, such as echo planar imaging (EPI), are applied. Moreover, the metabolite peaks in MRSI may be affected by spectral shifts and intravoxel broadening with potential peaks overlapping; also, the suppression of water and fat is more challenging due to widening of the water and fat spectral peaks. The implementation of higher order shimming has shown to significantly reduce the B₀ inhomogeneities, but unfortunately most human 7 T MR scanners are equipped with only second degree shims.¹²¹ Moreover, parallel imaging technique, such as generalized auto-calibrating partially parallel acquisitions (GRAPPA), in combination with readout-segmented EPI is able to reduce susceptibility-related distortions and improve the image quality thanks to the decrease of the acquired k-space data and a restricted field of view (FOV).¹²²

As the magnetic field strength increases, the Larmor wavelength for protons in the human head decreases and the RF wavelength in tissues becomes smaller than anatomical structures. The non-uniformity of B₁ derives from the interaction between the asymmetric and inhomogeneous human head, the RF coil, and the excitation sources.¹²³ This effect leads to changes in CNR and flip angles across the FOV, and a reduction of SNR from the center of the brain to the periphery and in the head-foot direction as well. Solutions to deal with B₁ inhomogeneities have been proposed. For instance, adiabatic RF pulses, modulating frequency and amplitude of the applied RF field above the adiabatic threshold, are relatively insensitive to B₁ inhomogeneities, allowing to uniformly rotate the net magnetization with a constant flip angle, and improving the outer-volume suppression.¹²⁴ Another solution is the employment of parallel excitation arrays with multiple independent transmit coils, in order to create a more uniform B₁ field by modelling the RF waveform and the RF pulse sequences on each specific channel.¹²⁵ B₁ non-uniformities represents a significant challenge not only for UHF imaging of the brain, but especially for spinal cord imaging, as the presence of vertebral bodies further exacerbates B₁ inhomogeneites. Despite many different solutions have been proposed, the optimal coil configuration for 7 T spinal cord imaging is not yet known and more efforts are needed to address this technical challenge.¹²⁶

SAR represents the measurement in watts per kilograms of RF power delivered during the scan to human tissues, with their subsequent heating. SAR increases roughly quadratically with the B₀, and other influencing factors include the total amount and frequency of RF pulses: sequences using large and very rapid RF pulses, such as spin-echo and TSE, lead to higher SAR, while GRE sequences (except TOF) typically result in lower SAR. FDA sets regulatory limits for RF safety and all MRI scanners are able to compute an estimate of the SAR before each acquisition in order to guarantee all subjects are safe during the scan. Methods to lower SAR include the use of parallel imaging techniques, the reduction of flip angles and number of echoes in multiecho acquisition, and the lengthening of TR. Longer TRs naturally result in longer acquisition times. This is still sometimes a challenge at 7 T.

Image artifacts such as ringing, ghosting, and blurring caused by motion of the subject during the MRI exam constitute a significant problem both for clinical 1.5 and 3 T MRI, and for 7 T MRI as well. Although imaging at UHF resulted in faster acquisition, the higher resolution derived from improved SNR lead to increased sensitivity to motion.¹²⁷ Even with co-operative subjects, the presence of physiological spontaneous movements, including breathing, heartbeating, and muscle relaxation, may cause motion artifacts.¹²⁸ Different motion correction procedures exist, including post-processing retrospective techniques and real-time prospective methods: the first is applied after the data have been acquired, while the latter approach detects the subject motion during the scan and dynamically adjusts the imaging protocol tracking the motion to reduce the artifacts. For instance, prospective motion correction system (Kineticor, HI), using an optical tracking system composed by a single camera and a moiré phase tracking marker to obtain head motion information in real time, has recently shown to be feasible and effective in reducing the artifacts caused by physiological movements at 7 T MRI.¹²⁹

Another limiting factor that impacts both research and clinical utilization of 7 T MRI is related to biomedical implants. Currently, relatively few implants have undergone proper testing at 7 T MR scanner, precluding a significant portion of subjects and patients from taking advantage of UHF MRI.^130–132

Moreover, although many 7 T protocols have demonstrated to provide images with a significantly higher level of details compared with 3 T MRI, increased resolution is often concomitant with increased image-encoding burden that can cause long scan times, a condition not always acceptable in clinical setting. For this reason, new acquisition methods, including RF pulse design schemes and parallel imaging approaches such as simultaneous multislice and volumetric 3D imaging, have been developed in order to obtain high-quality images over the entire brain within a relatively short time-frame.¹³³

Conclusion

7 T MRI is set to become a clinical tool in addition to numerous research applications. Several studies have demonstrated that the increased spatial resolution and SNR may have significant clinical implications for the diagnosis, treatment, and follow-up of patients in a number of neurological diseases. However, further studies comparing conventional, high-field, and UHF MRI are needed in order to verify on a larger scale the diagnostic and therapeutic advantages reported in previous studies discussed in this paper. Moreover, the development of new imaging, post-processing, and analyses techniques is desirable not only to overcome the technical challenges, but also to maximize the image quality attainable at UHF MRI so as to demonstrate current and discover future clinical applications.