Synopsis

MRI can be used to non-invasively encode and quantify different biophysical aspects of tissue including flow and motion. In this presentation, we will review methods that generate motion related image contrast, through manipulation of longitudinal magnetization prior to excitation, such as Time-of-flight MR Angiography, Arterial Spin Labeling, Myocardial Tagging; as well as through manipulation of transverse magnetization between excitation and acquisition, such as Phase Contrast MR Angiography, Diffusion Imaging, and MR Elastography.

Syllabus

MRI can be used to non-invasively encode and quantify different biophysical aspects of tissue including flow and motion.

A typical pulse sequence involves excitation, where longitudinal magnetization is converted to observable transverse magnetization using a combination of RF and gradient pulses; and data acquisition, where transverse magnetization is spatially encoded using additional imaging gradients for signal detection and image reconstruction. Gradients, used in conjunction with RF pulses to nutate magnetization, can be used to modulate longitudinal and/or transverse magnetization, by changing the local Larmor frequency and the phase of spins.

Methods for generating motion related image contrast, that can be categorized as follows, will be reviewed in this presentation:

A) Methods that modify Mz prior to excitation: These methods use a combination of RF and gradient pulses to modify the spatial distribution of longitudinal magnetization and observe the effects of the transport of these ‘tagged’ spins from one location to another. Since they are based on modulation of Mz, the time scale for these methods is limited by T1.

• Time-of-flight MR angiography: maximizes vessel conspicuity by manipulating imaging parameters in a conventional imaging sequence. When a fast gradient echo sequence is used with a large flip angle, the steady state magnetization in static spins in the imaging volume is low. However, blood flow continually transports relaxed blood spins into the imaging volume, therefore blood flowing in vessels tends to have larger longitudinal magnetization (and MR signal) than surrounding static tissue. Simple image processing methods such as Maximum Intensity Projection then serve to exclude non-flowing spins from the final image.

• Arterial spin labeling: differentiates inflowing arterial water from water in the tissue of interest by labeling the arterial water magnetization using spatially selective RF inversion (or saturation) pulses. A delay is inserted into the sequence to allow for labeled blood to reach the tissue of interest. Arrival of labeled blood in the tissue of interest decreases the tissue MRI signal in proportion to blood flow and other measurable parameters. Subtraction of labeled and control images (acquired without labeling) provides the perfusion sensitive signal, which with appropriate modeling assumptions (and additional data such as the T1 of tissue), can provide quantitative estimates of regional perfusion.

• Myocardial tagging: relies on temporary magnetic fiducial markers, or tags, which are generated by modulating the underlying image intensity using a specific presaturation (i.e. exciting multiple planes of magnetization, which for saturation tagging are then dephased) such that they make no significant contribution to subsequently acquired images (leaving null regions at tagged locations). After a certain time (i.e. a different phase of the cardiac cycle) another image is acquired where the tags are still visible. Deformation of the tags is analyzed to obtain a model of the underlying tissue motion. DENSE and HARP encode a uniform pattern of phase modulation into the tissue at a chosen time, and detect the deformation of that pattern at a later time to estimate local displacements.

B) Methods that manipulate Mxy (after excitation before data acquisition): These methods use only gradient pulses, i.e. motion sensitizing gradients. Contrast is based on phase shifts from motion along the gradient. A basic waveform is a bipolar gradient pulse, the first half of which generates a linear phase modulation in the direction of the gradient vector, and the second half reverses this modulation. If motion occurs during the pulse, the phase modulation is incompletely rewound, leaving a residual phase shift, proportional to the moving spins velocity, the strength of the applied gradient, and the square of the length of time it moves within that gradient. Since they use Mxy, the time scale for these methods is limited by T2 decay.

• Phase contrast MR angiography: can be used to quantify velocity of moving fluids such as blood and CSF. Bipolar gradients are typically added to a fast gradient recalled imaging sequence. A reference dataset corrects for other phase contributions that do not originate from motion. After subtraction of the reference phase from the phase image obtained with bipolar gradients, the phase of the image is directly proportional to the component of velocity that is along the direction of the applied gradient. Duration and magnitude of the bipolar gradient pair determines the velocity encoding (VENC). VENC is inversely related to the size of these gradients. For an optimal velocity to noise ratio, VENC should be set such that the highest velocity likely to be encountered within the vessel of interest corresponds to a 180° phase shift. Velocity aliasing occurs when VENC is set too low. Flow can be calculated by integrating the area-velocity product over an area of interest, e.g. a vessel.

• Diffusion imaging: provides information on tissue microstructure, similarly, using bipolar gradients to produce motion related phase shifts in individual spins. Since motion of spins due to diffusion is not coherent, there is a distribution of velocities (directions and/or magnitude), thus a distribution of phases (there is no net phase shift). Phase dispersion across spins means a less coherent signal with an attenuated magnitude. Diffusion weighting gradients typically need to be a few orders of magnitude larger than flow weighting gradients in order to generate measurable signal attenuation. The attenuation is related to the typical distances travelled by the diffusing spins (thus the diffusion coefficient D), the time between the diffusion weighting gradients, their amplitude and duration. Relevant experimental factors (i.e. gradient amplitudes and timings) are grouped together into the b-value (larger b-values induce more dephasing, hence more signal loss). Quantitative estimates of the apparent diffusivity along the direction of the applied motion-encoding gradient can be obtained from the ratio of the diffusion weighted intensities measured at different b-values.

• MR elastography: involves estimation of tissue stiffness through introduction of a known oscillation into the tissue of interest (i.e. acoustic shear waves applied with a mechanical driver) to generate standing waves in the tissue. The standing waves are imaged using the principles of phase contrast MRA, but because the displacement is periodic, periodic motion encoding gradients tuned to the vibration frequency can selectively sensitize the sequence to the applied vibrations while filtering out other sources of motion. The train of bipolar gradients is externally triggered to synchronized them with the external vibration, such that the shear waves generated in the tissue are encoded into the phase of the signal. Wave propagation, the local wavelength, and tissue deformation are then analyzed to estimate viscoelastic properties such as shear modulus.

Acknowledgements

References

Bernstein, King, Zhou. Handbook of MRI Pulse Sequences

Liang, Lauterbur. Principles of Magnetic Resonance Imaging

Nishimura. Principles of Magnetic Resonance Imaging

Haacke, Brown, Thompson, Venkatesan. Magnetic Resonance Imaging Physical Principles and Sequence Design