The trend forward high magnetic fields derives from the quite low sensitivity of MRI. This originates from the small difference in the population of two consecutive Zeeman energy levels caused by the difference in energy of these levels. The slight excess of spin in the fundamental state generates a small polarisation, which depends on the temperature and the magnetic field applied, and it is only from this small fraction of sample that the whole absorption signal is detected in correspondence with the radio frequency associated with the nuclear transition. One direct method for increasing the low sensitivity was identified as increasing the static magnetic field B₀. In more detail, the MR signal increases in quadratic mode with the strength of the static magnetic field, while the associated noise demonstrates linear dependence. Therefore, the use of MR 7 T equipment allows a significant increase in the signal-to-noise ratio (SNR) in comparison with lower fields. For example, an image obtained with a clinical 1.5 T or even a 3 T system has a spatial resolution limit of about 1 cubic millimetre, while the combination of the benefits of ultra-high field can lead to a resolution of some hundredths of microns with similar signal-to-noise ratio. The significant increase in SNR produces a great improvement of all the imaging parameters and can be used not only in terms of spatial resolution but also in terms of sensitivity to modifications of the composition of the tissue or in terms of temporal resolution for dynamic phenomena or in terms of spectral resolution of signals. In addition, many other benefits could come out from the potential of new sources of mechanisms of contrast. Particularly in the brain, but also generally in the human body, there are potential sources of signal which cannot be fully explored using standard magnetic fields, because of low SNR, poor spatial resolution and/or their scarcity. It is the case of myelin, iron or metabolites which contain nuclei that differ from hydrogen (¹³C, ²³Na, ³¹P) that can be in principle detected by MR. These further signal sources can be extremely useful for providing additional and complementary information about molecular structure and/or the physiological, metabolic and functional dynamics of physiopathological processes. Moreover, the improvement in SNR and consequently in the intrinsic sensitivity of MR experiment can make possible the study of spontaneous distribution of nuclei different from proton that are normally found within structures of biological interest. For the same reason, it becomes possible to explore the distribution and the metabolic dynamics of molecular probes that are artificially enriched with stable isotopes that are visible in MR.

Increasing the magnetic field however causes some physical and instrumental problems that must be considered and fully investigated so that the potential of an ultra-high field MR system can be fully exploited. The major issues are related to the signal losses associated with the effects of magnetic susceptibility and inhomogeneities of the static magnetic field B₀, to the increase of chemical shift artefacts and to the variations in the relaxation times T₁ and T₂, which completely change the semeiotics of the images and the strategies of signal acquisition.

Moreover, the wavelength and dielectric effects of the radio frequency (RF) signal produce an uneven excitation and can cause an inhomogeneous distribution of radio frequency energy deposition on tissues. At 7 T, the resonance frequency for hydrogen is 298 MHz. Raising the operating frequency leads to the so-called wavelength effect: the radio frequency wavelength becomes comparable with the dimensions of the sample being investigated (limbs, brain and trunk) which per se brings stationary wavelength effects to the whole sample. The experiment becomes more complicated due to the dielectric properties of the sample irradiated by the radio frequency that further decreases the radiation wavelength (e.g. at 7 T, the wavelength in anatomical tissue with a high concentration of water is about 12 cm). This effect shows up as inconsistencies in the RF field transmitted (B₁) distribution, which is known as dielectric resonance. This generates peculiar artefacts in images that present hypo- and hyperintense zones caused by the presence of peaks and dips in the RF magnetic field B₁ and as a consequence generates locally different flip angles. The inhomogeneities in the B₁ magnetic field inevitably translate into inhomogeneous depositing of energy on the patient and give rise to possible “hot spots” of energy deposited. Specific absorption rate (SAR) is the parameter that measures this energy and forms the basis for both national and international patient safety standards. Commercial UHF MR systems monitor this parameter and use acquisition thresholds to do not exceed the above-mentioned standards. However, given the uneven distribution of the transmitted fields and the possible presence of “hot spots”, new methods for assessing variations in local magnetic field are mandatory. These may involve the simulation of electromagnetic fields and experimental measures [1], as well as the development of sensors for the real-time monitoring of SAR specifically for each individual patient (Fig. 23.2).

In recent years, huge technological progress has been made in attempting to resolve the problem of inhomogeneities in the excitation and in the receiving of RF signals in terms of both hardware and software. Rewriting the pulse sequence for UHF to enable the use of less sensitive sequences and eliminate errors in the flip angles applied to prevent sequence “refocusing” has proved extremely difficult. As the use of multichannel coils for signal receiving revolutionised MRI by allowing increased SNR and reduction of acquisition times, the introduction of parallel transmission opened new scenarios for UHF, creating a fundamental turning point in the use of UHF. Parallel transmission uses and controls a multichannel array system for the signal transmission, adjusting independently phase and amplitude of signals sent to each channel, with the aim of producing uniform excitation inside the sample. This led to what is now known as “RF shimming” (radio frequency calibration). Recently, parallel transmission systems allow to manage the different channels of signal transmission not only by adjusting the phases and amplitudes of signals but also by sending completely independent signals to single channels, increasing the spatial selectivity of the RF pulses. Thanks to this technological advance, it is possible to compensate specific B₁ inhomogeneities induced by specific geometry and dielectric properties of each single human body exanimated. Corrections of B₁ are therefore “customised” and allow the quality of imaging to be improved while at the same time minimising the depositing of energy for each patient with his or her specific characteristics (gender, age, dimensions such as weight and height, muscle mass and the presence of physiopathological signs).

As described above, some physical key UHF MR phenomena are not favourable for the application of 7 T in a clinical context such as inhomogeneities in B1 and the corresponding distribution of SAR. Moreover, the shortening of the RF wavelength used at UHF can interfere more strongly with metal objects increasing the heating effect. For this reason, the presence of implanted metal objects remains the main cause of exclusion from UHF MR exams.

With knowledge of above-mentioned limits and the significant results expected from awareness of the physical phenomena of UHF MR, the encouraging results of experimental studies carried out on patients suffering from diseases of the central nervous system have recently started to arrive.