Course Summary

Microstructure is composed of an array of molecules other than water that support tissue structure and function. These molecules contain an abundance of hydrogen nuclei that could be measured by proton MRI, but their restricted motion makes their T₂’s far too short for direct detection with conventional imaging. Fortunately, much of this otherwise invisible proton pool can be detected indirectly because it exchanges with mobile water protons(1,2). This magnetization exchange occurs by spin diffusion along the molecule to sites exposed to water and then either chemical exchange (actual exchange of protons between water and the surface protons) or just magnetization transfer through spin-spin interactions(3,4). Magnetization exchange contributes to the apparent T₂ and T₁ relaxation of the mobile water pool, but can be further emphasized by selectively saturating the spins of these molecules with RF pulses. The exchange of magnetization transfers the saturation to the mobile water pool. Imaging of this saturation transfer from bound molecules is referred to as Magnetization Transfer (MT) imaging.

Selective saturation of the immobile molecular pool is achieved by exploiting its long T₂. On-resonance pulses that rotate the free water magnetization by an integer multiple of 180° can be used(5,6). Such pulses preserve the magnitude of the mobile water but mostly saturate the immobile proton pool because its T₂ is short compared to the RF pulse width. A more common approach is to apply RF pulses at relatively large frequency offsets relative to the water center frequency(1,2). Spins with long T₂ have very broad lines that extend many kHz away from center frequency. Off-resonance pulses saturate the very broad, short T₂ lines of immobile molecules while having a smaller effect on the narrow lines of mobile water. By acquiring images with different off-resonance frequencies, one can generate a z-spectrum of MT. Modeling of z-spectra can be used to characterize the T₂ and lineshape of the immobile molecules(7). Unlike the Lorentzian lines from spins in mobile liquids, lines from spins with restricted mobility are Gaussian, and when averaged across orientation angles, they are super-Lorentzian(8,9).

A number of strategies to quantify MT effects from images have been used(7). The simplest is to calculate the MT ratio, or MTR, defined as 1-(MT image)/(Reference image) where the reference image is an image acquired with no MT power applied. The MTR depends on the power and frequency of the MT saturation pulses applied but also on other details of the pulse sequence such as the repetition time, TR, the tissue T₁, and errors in RF amplitude caused by spatially nonuniform RF transmit fields and any miscalibration. A newer measure, MTsat, attempts to extract a measure of the effect of an individual MT pulse that is independent of T₁ and sequence parameters and also less sensitive to RF errors(10). Still, this measure cannot be directly related to any microstructure properties or components.

Fitting a full physical model for MT, a quantification process known as qMT, can be used to measure a number of immobile proton parameters including their fractional density relative to water, the exchange rate to water, and their lineshape(7). A full model fit requires acquiring a large number of RF saturation frequencies and powers, but the number of images needed can be reduced when the range of expected parameters is limited and when measurement of a particular quantity is targeted. In particular, approximate quantification of the immobile proton density relative to water, sometimes referred to as the Macromolecular Proton Fraction (MPF) can be achieved with as little as one off-resonance saturation(11).

More sensitive probes of microstructure could be possible if the total immobile proton pool can be divided into different microstructural elements. One promising new approach(12) aims to separate these elements based on a property of immobile proton lines, their dipolar order relaxation time (T_1d)(13). Dipolar order can be thought of as a distortion of the lineshape that occurs when RF is applied off-resonance. The line becomes more saturated on the side where the RF is applied while the opposite side is less saturated. This effect competes with T_1d relaxation. If T_1d is long, more distortion occurs and the saturation effect is reduced. This effect can be highlighted by subtracting an image acquired with power divided equally between positive and negative frequency offsets from a single frequency saturated image, an experiment known as inhomogeneous MT (ihMT) imaging(12). ihMT emphasizes longer T_1d molecules. T_1d’s appear longest in membrane lipids, so ihMT images emphasize membrane rich myelin. More refined ihMT experiments(14) can potentially measure other pools of molecules with intermediate T_1d’s.

MT has been used in a number of ways in research and clinical studies. MT is used in neuromelanin imaging to attenuate myelin signal and enhance contrast(15). The strong MT and ihMT effects in myelin can be used to provide quantitative metrics correlated with myelin to serve as biomarkers of disease for research and clinical purposes(16). Quantitative Myelin measures can be paired with diffusion measurements to estimate the g-ratio(17), the ratio of the axon radius to the myelin sheath radius.

Acknowledgements

No acknowledgement found.