In last two decades, the remarkable development of MR imaging has been accompanied by a marked increase in the field strength of imaging magnets, and we have witnessed a steady growth in field strength (over a 20-fold rise) from 0.015~0.3 Tesla to 3~7 Tesla on human MR systems ^[1]. Initially, ultra-high-field systems (beyond 3 Tesla) were designed only for small animal research. Recently, however, efforts have been undertaken to establish these systems for human brain and body imaging. These ultra-high-field human MR systems such as 7T MR have great potential to provide ever-advanced imaging capabilities for both clinical diagnosis and basic science research.

       This review article highlights some very exciting preliminary results of brain imaging recently obtained from 7T MR scanner and briefly discusses what can be expected in clinical applications from ultra-high-field human MR systems.

Ultra-High-Field MR and Challenges

       As the demands for better image quality and shorter acquisition time increase, high-field MR is becoming increasingly popular in the clinical setting, especially with recent considerable improvements in spatial encoding methods for fast acquisition and in radiofrequency (RF) engineering for development of state-of-the-art RF coils compatible with increased field strength ^[2]. There are several known inherent advantages associated with increasing field strength, including increased signal, enhanced susceptibility contrast, greater chemical shift, and increased blood oxygenation level-dependent (BOLD) effects. Disadvantages of high magnetic field strength include increased B0 and B1 inhomogeneity, as well as blunting of certain forms of relaxation-related contrast. We discuss briefly here the strengths and challenges that are related to these changes at ultra-high-field MR.

It is all about signal

       Signal intensity, the fundamental quantity underlying MR image resolution and contrast, is proportional to the square of the static magnetic field strength (B02) because both the rate of proton spin precession and the voltage induced in RF detectors by precessing spins increase linearly as the magnetic field increases ^[3]. On the other hand, common sources of noise received by RF detectors only increase linearly with B0, resulting in a predicted linear increase in signal-to-noise ratio (SNR) with field strength. Compared with the clinical standard field of 1.5T, a 7T MR imaging system boosts the SNR by a factor of approximately 5. Such an increased SNR will have numerous benefits in improving the image quality and clinical utility of both routine and advanced MR imaging. First, ultra high spatial resolution can be achieved with increased SNR, since subdivision of an image into progressively smaller voxels is known to exact a high cost in SNR. High spatial resolution is critical for medical and scientific discoveries in live human brains within a reasonable scan time. Second, the increase in SNR is important for advanced functional MRI applications (i.e. BOLD). Functional signal changes can become more significant at high field due to both to the increased achievable spatial resolution and to the increased susceptibility effects. Third, 3D high resolution susceptibility-weighted imaging (SWI) becomes a more promising sequence at high field with the increase of SNR, providing a better resolution in visualizing small veins and micro bleeding. Fourth, arterial spin labeling (ASL) techniques for cerebral blood perfusion measurement are also expected to benefit from increased SNR together with increased blood T1 relaxation times at high field.

       Additionally, with substantial increases of SNR at ultra-high field, the capability of MR spectroscopy (MRS) to detect and distinguish small quantities of metabolites in the brain can be significantly improved^[4]. The increased spectral resolution of ultra-high-field MRS will improve the accuracy and localization of specific metabolites including GABA, glutamine and glutamate that are unavailable on conventional field MR, and the increased SNR will enable both spectroscopy and imaging of nuclei other than hydrogen, such as phosphorous, sodium, and carbon. Therefore, 7T MR promises to enable a wide range of applications for non-invasive anatomical, functional, and metabolic investigations.

New horizons in image contrast

       As the main magnetic field increases from 1.5T to 7.0T not only is there a remarkable increase of SNR, but there is also a clearly notable change in image contrast. This is because T1 relaxation times for various tissue types increase and converge, whereas T2 and T2^* are reduced. The most prominent change of image contrast is that susceptibility or T2^* effects are significantly boosted at ultra-high field strength. The increased susceptibility contrast at high fields offers important benefits for many sequences including SWI which exploit paramagnetic contrast and susceptibility differences. SWI applies phase information to multiply magnitude data, maximizing the susceptibility effect due to any cause of local field inhomogeneity including venous deoxyhemoglobin and brain iron accumulation. SWI venography, using venous deoxyhemoglobin as an intrinsic T2^* contrast agent, plays an important role in detecting vascular diseases such as venous malformation and cavernous angioma. The visualization of microvascularity in these diseases and in brain tumors has been reported to increase at high field ^[5].

       BOLD sequences can also benefit from the boosted T2^* effect, since the signal changes on BOLD are derived from the fact that the local concentration of deoxyhemoglobin changes during neuronal activity. There is only 0~3% BOLD signal changes at 1.5T, whereas these changes can be up to 6% at 7T ^[6]. For dynamic susceptibility contrast-enhanced perfusion MR imaging, reduced doses of contrast agent may be achievable at higher field. By effusively utilizing the enhanced susceptibility effects, the unique image contrast achieved in these sequences may provide new insights into a number of diseases, particularly those with venous structure involvement (e.g. multiple sclerosis), angiopathy, subtle hemorrhage, and abnormal iron deposition.

Technical challenges

       Although ultra-high field strengths provide many advantages in brain imaging, problems and challenges also arise at such high magnetic fields. First, at ultra-high-field MR, interaction between the human brain and the RF field can produce images with inhomogeneities-most commonly a center brightening-due to the dielectric properties of brain tissue. Second, the enhanced susceptibility effects with signal intensity loss and distortion near the skull base and air sinuses become prominent at higher field strengths, which can substantially affect the image quality. Third, due to prolonged T1 relaxation time at high field, image contrast may be altered and conventional T1-weighted imaging may be less effective than at lower field. However, MPRAGE, a high resolution sequence with T1-weighted properties, may be an alternative for T1 imaging. Fourth, there is a major concern relating to increased power deposition or specific absorption rate (SAR) when ultra-high magnetic field MRI is performed in humans ^[7]. RF energy is absorbed more effectively at higher frequencies, and safeguards implemented to prevent tissue heating may limit pulse sequence performance at ultra-high field strength. Lastly, with the increase in field strength, design and construction of effective RF transmitter and detector coils becomes more challenging. This difficulty arises in part from the electrical coupling between RF coils and the human brain.

       The challenges mentioned above complicate the transition from current clinical routine field strengths to ultra-high field strengths for human applications. However, significant efforts are currently being devoted to addressing these problems, including research into the use of parallel transmission and detection.

Our Initial Results of Brain Imaging

       Over the past few years, there are increasing numbers of advanced imaging centers in major academic institutes that have installed human 7T MR systems. The 7T magnet is the most powerful approved whole body human MR machine in the US. Our 7T (Siemens MAGNETOM) MR at New York University Langone Medical Center uses gradients of 45 mT/m with maximum gradient strength of 72 mT/m effective and a slew rate of 200 T/m/s (346 T/m/s effective).

       For brain imaging, ultra-high-field MR systems have great potential to detect otherwise invisible structural and physiological abnormalities. These ultra-high-field strength machines are incredibly important to the future of our understanding of how the brain works, and they will ultimately help us find answers to some of the most challenging questions that face the medical profession. Just recently, we have performed a series of experiments to establish and optimize methods for brain imaging at 7T using a 24-element head coil array (Nova Medical, Inc. in Massachusetts). The coil array is comprised of two separate components: a birdcage like circularly polarized transmit coil and a 24 element phased array positioned on a close-fitting helmet-like former to maximize signal to noise ratio. The coil plugs directly into the patient table of the MAGNETOM 7T scanner. We have acquired a series of volunteer data, data in patients with multiple sclerosis (MS), and postmortem data in brain tissue samples from patients with Alzheimer's disease.

       One of our efforts has involved optimizing imaging parameters to attain fine anatomical imaging with high resolution and contrast. With careful shimming and selection of imaging parameters, signal gains on the 7T can be used to achieve excellent image resolution. The unique image contrast due to enhanced T2^* effect adds better characterization of venous architecture and brain iron assessment. As shown in Figure 1, high-resolution anatomical images not only depict exquisite details of small structures, but also offer excellent tissue contrast particularly in regions with iron-rich neuron nuclei, cortical layers, and white matter tracts. On SWI venography, the enhanced susceptibility effects from deoxyhemoglobin allow enhanced visualization of small veins. A remarkable example is shown in Figure 2: even with a thin image slice (1 mm) and very high resolution (pixel size: 200 µm×200 µm), the image SNR is still high enough to produce exceptionally fine venography. Such high resolution in vivo MR venography, with quality superior to that of the invasive venous angiography used clinically, has many clinical applications in diseases involving venous structures. The microvenules, which are shown on SWI high resolution phase images (Figure 3), have never been so noticeable at conventional field strengths. This is critical in functional MRI (i.e. BOLD) studies for improved identification of neuronal activity, since at low-field it is difficult to differentiate whether the BOLD signal changes are from micro cortical veins or from activated parenchyma.

       Recent work of my multiple sclerosis (MS) has been focused on elucidating the pathogenesis of the disease, which is proposed to be primarily related to microvascular abnormalities (a consequence of the perivenous inflammation known to occur in MS), leading to the primary and initial insult to myelin and axons. Although the role of vascular pathology in MS was suggested long ago by histopathological studies, such perivenous relationship of MS lesions has not been identified in vivo prior to recent advances in MR. Ultra-high-field MR offers significant benefit in directly revealing the close relationship between venous injury and lesion development in the very early stages. We have evaluated this advantage on 7T in MS patients. As shown in Figure 4, virtually every MS lesion identified by MR susceptibility sensitive imaging reveals a central vein, indicating the microvascular pathogenesis of MS ^[8]. More importantly, using this high resolution susceptibility sensitive imaging on 7T, we can clearly see many subtle abnormal changes of microvascular wall and perivenous inflammatory abnormalities that have never been identified at conventional field strength. These vascular abnormalities at the pre-lesional stage or initial stage of lesion formation have important implications not only in directly unveiling the mirovascular pathogenesis, but also in developing the new targets to intervene pharmacologically in the early stages of MS. Today, MR imaging is the most important paraclinical tool for MS ^[9]. These novel findings of remarkably increased detection of early vascular abnormalities at 7T will have direct impact on our understanding of early disease activity, early treatment, and therapeutic monitoring in MS.

       In addition, due to the high spatial resolution and improved contrast at high field, small cortical MS lesions which are usually documented only in histopathological studies can now be visualized at 7T (Figure 5). Since the mechanism of lesion formation in gray matter differs from that for lesions in white matter, the intimate relationship between lesions and veins in white matter may not be seen in gray matter. Finally, the detection of MS lesions on T2-weighted imaging is significantly improved at high field strength.

       In recent years, there is an emerging interest in detecting the iron content of amyloid plaques in Alzheimer's disease (AD) at ultra-high-field MR. Any imaging technique capable of directly visualizing amyloid plaques - the hallmark of AD - will be clinically important for early and accurate diagnosis of AD. Intriguing data from animal studies have demonstrated the possibility of detecting iron-induced susceptibility effects in plaques using T2^*-weighted imaging at ultra-high field. We have obtained some postmortem AD brain samples at New York University Department of Pathology. These data are also exciting. As shown in Figure 6, compared with age-matched control samples, numerous plaque-like hypointensities in the frontal gyrus can be visualized in the AD sample on T2^*-weighted imaging at 7T. These small hypointensities detected on ultra-high-field MR can be matched with histopathological evaluations. However, there is no in vivo data from live human brain showing similar findings, probably because the plaques are too small (usually 20~160 µm) and the iron-induced susceptibility is not strong enough to be detected at conventional field strength. Due to enhanced susceptibility effects at ultra-high-field, the actual size of these plaques may be smaller than what is visualized on the sensitive susceptibility-weighted imaging. This suggests that noninvasive direct detection and quantification of amyloid plaques in vivo in live human brain may become feasible in the near future based on ultra-high-field MR. In addition, high resolution imaging has the potential to demonstrate the detailed inner structures of the hippocampus and atrophic changes in AD specimens (Figure 7)^[10].

       In summary, preliminary data obtained at 7T has revealed the great potential - both for basic understanding and for clinical diagnosis - of ultra-high field MR imaging.

Click for larger

Fig 1.  Representative images (acquired in less than 7 minutes) from a normal volunteer scanned with susceptibility sensitive imaging on 7T demonstrated a unique image contrast of iron deposition in the ferruginated (iron rich) neurons in basal ganglia and brain stem (thick arrows), venous structures (long thin arrow) due to high concentration deoxyhemoglobin, and white matter tracts such as optic radiation (short thin arrows). There was also a significant improvement in visualizing the detailed cortical structures with high resolution imaging (arrowheads).

Click for larger

Fig 2.  High resolution SWI venography from a normal volunteer showed extremely fine cerebral venous anatomical imaging. These were SWI minimum intensity projection (mIP) images with pixel size of 0.2 mm×0.2 mm with 8 mm brain coverage. Such high resolution venography without contrast agent could only be achieved at ultra-high-field strength due to its considerably enhanced susceptibility effect of venous deoxyhemoglobin and superior conspicuity due to sufficient signal.

Click for larger

Fig 3.  SWI phase images with 0.2 mm×0.2 mm×1 mm voxels from a normal volunteer demonstrate very small cortical veins, which were not seen with conventional field MR. The recognition of such tiny venous structures in the cortex had important implications for BOLD functional MRI.

Click for larger

Fig 4.  High resolution susceptibility sensitive imaging with 0.2 mm×0.2 mm×1 mm voxels from a MS patient (55 years old, female) showed a close relationship between MS lesions and centered small veins (arrows), suggesting a primary vascular pathogenesis of MS lesions. Note many lesions were identified in the very early stage when the microvascular wall and perivenous space were just beginning to be affected (grey arrows). These subtle vascular abnormalities in the initial pre-lesional stage, to our knowledge, were the first time to be demonstrated in vivo and had never been shown on conventional field strength. These novel findings had important implications for early and new targets to intervene pharmacologically in MS.

Click for larger

Fig 5.  High resolution susceptibility sensitive imaging with 0.2 mm×0.2 mm×2.0 mm voxels from a MS patient showed many tiny lesions in the cortex (arrows). The ultra-high-field MR provided a powerful view of the cerebral cortex, allowing a largely improved distinction of cortical layers and identification of small signal changes in the cortical gray matter. Note that image resolution and contrast remained even on the largely magnified image because of the plenty signal provided at 7T.

Click for larger

Fig 6.  Optimized three-dimensional susceptibility-weighted imaging (SWI) was performed with an isotropic resolution of 200 μm (or voxel size: 0.008 mm³) and acquisition time of about 53 minutes demonstrating that many more plaque-like hypointensities in the AD sample as compared with the age-matched control sample. With largely enhanced susceptibility effects and superb signal on 7T, amyloid plaques, which contain iron deposition, can probably be directly visualized in vivo.

Click for larger

Fig 7.  As expected, in addition to excellent image contrast, 7T imaging also offers high spatial resolution with a sufficient enough signal to show the details of small structure such as the hippocampus in AD. The signal and atrophic changes in CA1 to CA3 are far more appreciated on 7T than conventional field strength.

Future Directions - The Decade of Ultra-High-Field MR

       Although significant technical advances are still required and ultra-high-field MR research is still in its early stages, recent important advances in coil technology, parallel imaging, and state-of-the-art software for ultra-high-field strengths promise to bring the advantages of ultra-high-field MRI closer to the clinic. At field strengths as high as 7T, new areas of research are opening up in microscopic imaging, molecular imaging, high resolution/contrast vascular imaging, biochemical imaging, functional brain imaging, and RF technology. The advanced capabilities for brain imaging afforded by ultra-high-field MR will be exploited in many CNS diseases such as multiple sclerosis, head trauma, brain tumor, epilepsy, ischemia, Parkinson's disease, and Alzheimer's disease. Recently, in recognition of this promise, orders for and installations of ultra-high-field magnets have become more common than ever world-wide. In the next decade, ultra-high-field systems will evolve into a new standard for clinical care and scientific discovery.