Home
This Title All WIREs
WIREs RSS Feed
How to cite this WIREs title:
WIREs Syst Biol Med
Impact Factor: 3.542

MRI in systems medicine

Full article on Wiley Online Library:   HTML PDF

Can't access this content? Tell your librarian.

Abstract Magnetic resonance imaging (MRI) is one of the primary medical imaging modalities and a key component of the standard of care in modern healthcare systems. One of the factors that distinguishes MRI from other imaging methods is the ability to program the MRI system to reveal a wide range of imaging contrasts, where each type of contrast offers unique information about the biological sample of interest. This ability stems from the fact that both the amplitude and phase of the magnetization of the underlying tissue can be manipulated to highlight different biological phenomenon. The flexibility and capabilities offered by modern MRI systems have enabled the development of a myriad of techniques for characterizing anatomy, physiology, and function. These include methods to characterize gross anatomy, tissue microstructure, bulk blood flow, tissue perfusion, and functional changes in blood oxygenation. This article is categorized under: Laboratory Methods and Technologies > Imaging Translational, Genomic, and Systems Medicine > Diagnostic Methods
Key elements of an MRI experiment. (a) In the absence of an external magnetic field, there is no preferred direction for the spins and so the net magnetization is zero, as shown on the left. When the sample is placed in the presence of an external magnetic field B0 there is a tendency for the spins to align with the external field to create net magnetization that is aligned with the field, as shown on the right. (b) An RF pulse is used to tip the magnetization away from the external field. In this example, the magnetization is tipped into the transverse plane using an RF pulse with a 90° flip angle. (c) The tipped magnetization precesses around the main field and also exhibits both transverse and longitudinal relaxation. In the top plot, the magnetization starts off at Mx = 0, My = 1, Mz = 0 after the application of the RF pulse (where the magnetization values have been normalized by M0 such that 1 denotes M0). The magnetization precesses around the z‐axis while at the same time exhibiting both T1 and T2 relaxation. The blue line in the lower left‐hand plot depicts the return to equilibrium of the longitudinal component Mz of the magnetization with a time constant of T1. The red line in the plot indicates the value of the equilibrium magnetization M0. The upper envelope (dotted green line) in the lower right‐hand plot shows the magnitude of the transverse component decaying with a time constant of T2. The lower envelope is provided to aid with the visualization of the overall envelope of the transverse component. The oscillating and decaying solid blue line in this plot depicts the My component. (d) Gradient fields are used to create spatial variations in magnetic field. Spins in regions with higher magnetic field will precess at a higher frequency as compared to spins in regions with a lower magnetic field
[ Normal View | Magnified View ]
(a) An arterial spin labeling (ASL) MRI experiment typically consists of tag and control sections. In each tag section, spins in the feeding arteries undergo magnetic inversion and then flow into the imaging region while undergoing longitudinal relaxation. After waiting for the spins to arrive in the imaging region, an image is acquired before the spins have fully relaxed. In each control section, fully relaxed spins flow into the imaging region and an image is acquired. The difference of the control and tag images is proportional to cerebral blood flow. (b) Cerebral blood flow maps are generated by taking the average difference of the control and tag images. (c) In most ASL experiments, multiple control and tag images are acquired in an interleaved fashion. Temporal changes in perfusion can be estimated by taking the running difference of the control and tag images. The blue curve shows the perfusion response (arbitrary units) to a visual stimulus, where the time period during which the stimulus was applied is indicated by the red bar
[ Normal View | Magnified View ]
(a) Time‐varying gradients can be used to assess diffusivity. In this example, both spins start off at the same position along the x‐axis and then move randomly in the presence of the gradient field. In the first half of the experiment (t ∈ [0,T]), a negative gradient is applied such that both spins experience a negative field offset that causes them to precess more slowly (and accrue positive phase since the precession is clockwise) than a spin located at the origin. The sign of the gradient is flipped for the second half of the experiment (t ∈ [T, 2T]). Due to the random motion of the spins, the positive field offset during the second half of the experiment does not cancel out for the effects of the negative field offset during the first half. As a result, both spins have acquired a net phase. If we imagine a collection of many spins diffusing, then there will be a distribution of phases at the end of the experiment, where the width of the distribution increases with diffusivity. (b) The top row show images with no additional diffusion weighting (b = 0) and substantial diffusion weighting (b = 3,000), where the direction of the diffusion gradients for the b = 3,000 image is largely perpendicular to the displayed image slice. As a result, the brighter parts of the image indicate regions where the primary direction of diffusion lies within the image plane. By combining measures from images acquired with different diffusion gradient directions, it is possible to obtain a map of the mean diffusivity (lower left), where the brightness of the image is proportional to diffusivity. A map of fractional anisotropy can also be obtained (lower right), where the colors red, green, and blue indicate fibers running in the left–right, anterior–posterior, and superior–inferior directions, respectively, and the brightness indicates the degree of anisotropy of the diffusion tensor at each location. (c) Estimates of the primary direction of diffusion at each location can be combined to form maps of the white matter fibers
[ Normal View | Magnified View ]
(a) Signal dephasing and the presence of magnetic field inhomogeneities. At time t = 0, the spins are in phase and the resulting MRI signal (vector sum of the spins) is indicated by the blue arrow. In the presence of moderate spatial inhomogeneities in the magnetic field (upper diagram), the spins experience slightly different fields and precess at different frequencies. Thus at the measurement time t = TE, there will moderate dephasing of the spins leading to a decrease in the MRI signal, as indicated by the shorter blue arrow. In the lower half of the panel, the degree of field inhomogeneity is much greater and so there will be more dephasing of the spins. This leads to a more pronounced decrease in the MRI signal. (b) In this example, the subject tapped the fingers of his right hand. This resulted in blood oxygenation level dependent (BOLD) signal changes in the left motor cortex (indicated by the yellow colors). Representative BOLD time courses from this region are shown on the left. Finger tapping leads to an increase in blood oxygenation, which leads to a decrease in signal dephasing and an increase in the MRI signal
[ Normal View | Magnified View ]
(a) Time‐varying gradients can be used to measure the velocity of moving spins. The black spin is stationary while the red spin is moving in the + x direction with a constant velocity v. In the first half of the experiment (t ∈ [0,T]), a negative gradient is applied such that both spins experience a negative field offset that causes them to precess more slowly (and accrue positive phase since the precession is clockwise) than a spin located at the origin. The moving spin flows into a region where the magnetic field grows progressively weaker and thus precesses even more slowly than the stationary spin. At time T the stationary and moving spins have phases of 90 and 135°, respectively. The sign of the gradient is reversed for the second half of the experiment (t ∈ [T, 2T]). For the stationary spin, the effects of the positive field offset during the second half of the experiment exactly cancels out the effects of the negative field offset during the first half so that it has zero phase at the end of the experiment. In contrast, the moving spin keeps moving into a region with increasingly stronger magnetic fields. The effects of the positive field offset during the second half of the experiment overcompensate for the effects of the negative field offset during the first half, such that the spin has accrued a net phase of −45° at the end of the experiment. This phase is directly proportional to its velocity. (b) The phase contrast angiography (PCA) image in the top plot depicts flow towards the brain through the internal carotid and vertebral arteries, indicated with the black shading. Flow away from the brain through the internal jugular veins is indicated with the white shading. The time‐varying flow velocities during the initial portion of the cardiac cycle are shown in the bottom plot for the left and right carotid (LCA, RCA) and vertebral (LVA, RVA) arteries
[ Normal View | Magnified View ]
Relation between extent of k‐space coverage and spatial resolution of the reconstructed images. In the left column, the depicted Fourier transforms differ in the extent of k‐space that is covered. The areas of k‐space covered in the first and second rows are 6.25% and 25%, respectively, of the area covered in the third row. The central portions of k‐space contain information from low spatial frequencies, whereas the outer parts of k‐space contain information from higher spatial frequencies. The images reconstructed from each of the Fourier transforms are shown in the right column. As the k‐space coverage increases to include higher spatial frequencies, the reconstructed images include finer spatial details and thus exhibit higher spatial resolution
[ Normal View | Magnified View ]
In a basic MRI study, the acquired information is the Fourier transform M(kx,ky) of the object m(x,y), where m(x,y) denotes the transverse magnetization as a function of spatial location. To acquire this data, linear gradient fields are used to cause spins at different spatial locations to precess at different frequencies. The spatial variation in precession frequencies results in an associated spatial variation in the phases of the spins in the object. In the lower left‐hand plot, a magnetic field gradient ΔBz(x) along the x‐axis produces a spatial variation in phase along the x‐direction. This phase variation is represented mathematically using the phasor notation exp(−j2π(kxx + kyy)), where kx and ky denote the spatial frequencies in the x and y directions, respectively. In this example ky = 0 as there is no spatial variation of the phases in the y direction. Because the phase variation is imposed upon the transverse magnetization of the object, the encoding process can be represented as the product m(x, y)exp(−j2π(kxx + kyy)) of the object and the phase variation. Integration of this product over space results in the Fourier transform M(kx,ky) of the object at the spatial frequencies kx and ky. The integration is performed implicitly by the RF coil, which is sensitive to the magnetization from the entire volume (depicted as the yellow ellipsoid within the coil). The green dot in the upper right‐hand image indicates the point in k‐space for which information is acquired using the phase variation shown in the lower left‐hand plot. Over the course of an MRI scan, the information from many points in k‐space is acquired. This is accomplished by manipulating the timing and amplitude of the gradients to impose different phase variations on the object (corresponding to different values of kx and ky) such that the acquisition process gradually fills in the values of the Fourier transform at different points in k‐space. At the completion of the scan, an inverse Fourier transform is used to convert the Fourier transform samples into a representation of the object
[ Normal View | Magnified View ]
Formation of a spin echo. At the start of the experiment, the net magnetization (solid blue arrow) is fully aligned along the y‐axis. Due to inhomogeneities in the magnetic field, spins will precess at different frequencies and acquire different phases. In this example, the green arrow in the upper middle plot depicts a spin that precesses at a higher frequency than the blue spin and has acquired a negative phase Δϕ(TE/2) = −ϕ0 relative to the phase of the blue spin. The red arrow depicts a spin that precesses at a lower frequency and has acquired a positive relative phase Δϕ(TE/2) = +ϕ0. The application of the 180° RF pulse inverts the signs of the relative phases of the spins such that in the lower middle plot the green spin now has a positive relative phase Δϕ(TE(+)/2) = ϕ0 and the red spin has a negative relative phase Δϕ(TE(+)/2) =  − ϕ0. (Because it flips the relative phases of the spins, the action of the 180° RF pulse is sometimes referred to as a “pancake flip”). In the period after the application of the 180° RF pulse, the green spin will continue to precess at a higher frequency, and will thus resume accruing a negative relative phase, such that at time t = TE its relative phase will return to zero (i.e., Δϕ(TE) = Δϕ(TE(+)/2) − ϕ0 = 0). In a similar fashion, the red spin will resume accruing a positive relative phase such that its phase Δϕ(TE) = Δϕ(TE(+)/2) + ϕ0 = 0 also returns to zero at t = TE. Thus, at time TE, the spins have been refocused and a spin echo has been formed
[ Normal View | Magnified View ]
(a) Spin‐echo pulse sequence. In the top row, the spacing between the 90° RF pulses is called the repetition time (TR). After the application of each 90° RF pulse, the magnetization is flipped from the longitudinal axis onto the transverse plane, such that Mz = 0. The longitudinal component then recovers exponentially and attains a value of Mz(0) = M0(1 − exp(−TR/T1)) just prior to the next 90° pulse. After each 90° pulse, the transverse magnetization starts out with this value and then exponentially decays with a time constant of T2, such that Mxy(t) = Mz(0)exp(−t/T2). The contrast of the sequence is evaluated at the echo time (TE). In this sequence a 180° refocusing pulse is used to create a spin echo. (b) Different types of image contrast can be achieved through the choice of values for TR and TE. The table lists typical proton density, T1, and T2 values for gray matter, white matter, and cerebrospinal fluid (CSF) at a field strength of 3 Tesla, where the proton density values are normalized by the CSF values. For the T1‐weighted image, the use of a short TE minimizes the sensitivity to T2 and the choice of TR to lie halfway between the T1 values of gray and white matter creates contrast between these two tissues types based on differences in T1. For the density‐weighted image, the use of a long TR and short TE makes the image relatively insensitive to T1 and T2, such that the resulting contrast largely reflects differences in proton density. For the T2‐weighted image, the use of a long TR and the choice of TE to lie halfway between the T2 values of gray and white matter creates contrast between these two tissues types based on differences in T2
[ Normal View | Magnified View ]

Browse by Topic

Laboratory Methods and Technologies > Imaging
Translational, Genomic, and Systems Medicine > Diagnostic Methods

Access to this WIREs title is by subscription only.

Recommend to Your
Librarian Now!

The latest WIREs articles in your inbox

Sign Up for Article Alerts