Synopsis

Keywords: Contrast mechanisms: Electromagnetic tissue properties, Contrast mechanisms: Relaxometry, Image acquisition: Sequences

This educational talk will review the principles underpinning several sources of contrast in MR images. Learning objectives are as follows:

• Understand the basic molecular mechanisms underlying image contrast
• Understand why various pulse sequence parameters influence contrast
• Understand how additional pulse sequence components can be used to generate different contrast

Syllabus

In addition to the use of non-ionizing radiation, the myriad ways to generate soft tissue contrast with MRI set it apart from other imaging modalities. This talk will explore several ways that contrast is generated with MRI by reviewing the molecular mechanisms underlying all MR imaging and then exploring how additional pulse sequence components can be used to generate different contrast.

The two molecular mechanisms underlying every MR experiment, and involved in the generation of contrast in most MR images, are T1 and T2 relaxation. These two independent, but coincidental, phenomena account for the return to thermal equilibrium of the net nuclear magnetization after the MR signal is generated. T1, or spin-lattice, relaxation reflects regrowth of the net magnetization along the longitudinal axis (M_z), while T2, or spin-spin, relaxation is characterized by dephasing of the transverse components of the net magnetization (M_xy).

T1 relaxation requires a stimulated energy transfer from nuclear spins to their surrounding environment, or lattice, by another magnetic field fluctuating near the Larmor frequency (ω₀). To be efficient at T1, a molecule must be rotating near ω₀, like that of fat tissue with a relatively short T1 (~250ms at 1.5T), as opposed to free water, which is small and tumbles too rapidly for efficient energy transfer and thus has a long T1 (~4000ms at 1.5T).

T2 relaxation may occur with or without a transfer of energy, and is thus caused by anything that also causes T1 relaxation, but can also occur without T1. The primary causes of T2 relaxation include dephasing by local magnetic field disturbances and flip-flop exchanges between spins. Additionally, inhomogeneities in the local magnetic field, either from intrinsic defects in the static magnetic field (B₀) or susceptibility-induced distortions, cause further transverse dephasing, which is a process termed T2*.

The longitudinal relaxation time, T1, represents the time required for M_z to regrow (1 - 1/e), or ~63%, of its final value. Conversely, the transverse relaxation time, T2, represents the time required for M_xy to decay to (1/e), or ~37%, of its initial value. Generally speaking, T1 is always longer than T2, which is always longer than T2*, or T1 > T2 > T2*.

The inherent T1 and T2 properties of various tissues can be exploited to highlight or minimize those tissues and create specific contrast in the resultant image by carefully timing certain elements in a pulse sequence. For example, in spin echo (SE) sequences at 1.5T, a short TR (~400-600 ms) maximizes T1 effects and a long TR (~2000-4000 ms) minimizes T1, while a short TE (~3-10ms) minimizes T2 effects and a long TE (~80-120ms) maximizes T2. Thus, a T1-weighted image can be generated by utilizing a short TR and short TE, while a T2-weighted image utilizes a long TR and long TE, and finally, a proton density, or PD-weighted image utilizes a long TR and short TE. Note, just because a sequence has a particular weighting does not mean that the other relaxation processes do not contribute to the image contrast. For example, proton density contrast is always present and represents ~30-50% of the contrast effect on both “T1-weighted” and “T2-weighted” images; similarly, “T1-weighted” images often have as much contribution from T2 effects as from T1 effects.

Gradient recalled echo (GRE), or simply gradient echo, images can also achieve different contrast weighting by manipulating TR, TE, and flip angle (FA). For example, at 1.5T, one can use a short TE (~4-8ms) to minimize T2* effects or a long TE (~15-20ms) to maximize T2* effects, and combining a large FA (~30-50°) with a short TR (~5-10ms) to maximize T1 effects or a small FA (~5-20°) with a long TR (~200-600ms) to minimize T1 effects. These timing schemes can be used to generate T1-, PD-, and T2*-weighted images; T2* is noted here since there is no 180° RF pulse, like in spin echo, to refocus the inhomogeneities that yield T2* relaxation. GRE pulse sequences use regularly-spaces RF pulses, with FAs typically less than 90°. Each RF pulse generates a free-induction decay (FID) signal; each subsequent RF pulse (from TR to TR) creates a spin echo; finally, each set of three RF pulses generates a stimulated echo. Any or all of these FIDs and echoes can either be spoiled, using RF or gradient spoiling, or refocused to generate a variety of different types of contrast.

Gadolinium-based contrast agents (GBCAs) can be injected to highlight various tissues in the body. Gadolinium is paramagnetic, so it facilitates both T1 and T2 relaxation, but with typical extracellular concentrations, T1 effects dominate. Thus, T1-W imaging can be performed pre- and post-contrast to accentuate what tissues uptake the contrast. Dynamic contrast-enhanced (DCE) imaging can be performed by rapidly acquiring a series of T1-W images to observe the wash-in and wash-out of the contrast, which is useful in delineating tumors and cardiovascular lesions. Furthermore, the T2 relaxivity properties of GBCAs can be captured with dynamic susceptibility contrast (DSC) perfusion imaging with a rapid series of T2*-W images.

In addition to these image weighting techniques, various components can be added to a pulse sequence in a modular fashion to emphasize or de-emphasize certain characteristics of the MR signal and phenomena in the body.

Fat or water can be suppressed using a few techniques. Chemical shift selective (CHESS) fat saturation utilizes the fact that fat and water have a chemical shift of 3.5 ppm (~220Hz and 1.5T and ~440Hz at 3T) by adding an RF pulse tuned to fat, which when combined with spoiler gradients, saturates and dephases the fat spins, leaving only the water spins for imaging. Dixon is a multi-echo technique to generate in-phase and opposed-phase images based on the difference in resonance frequency between water and fat, which can be combined in an additive or subtractive way to generate water-only or fat-only images. Inversion pulses can also be used to exploit differences in T1; STIR uses a short inversion time or TI (~150-180ms at 1.5T) to null any signal with a short T1 including fat; FLAIR uses a long TI (~2000-2500ms at 1.5T) to null water from CSF with a long T1.

When a proton moves by diffusion throughout an MR experiment, it experiences an unrecoverable phase shift as it travels through different intrinsic and extrinsic magnetic fields. This is a pure T2 effect which is proportional to the square of the gradient field and the length of time the spin moves within that gradient. Diffusion-weighted imaging (DWI) measures the average diffusion of spins within a voxel by encoding the phase shift into the MR signal with two diffusion-sensitizing gradients. If a minimum of six diffusion directions are acquired, one can perform tractography using diffusion tensor imaging (DTI).

MR spectroscopy (MRS) is a powerful tool to identify and quantify metabolites in the body by taking a spectrum of the MR signal. Since water is the dominant signal in the human body, followed by fat, the suppression of both of these signals is critical for MRS. Both single-voxel and multi-voxel techniques can be utilized to visualize brain metabolites like N-acetyl aspartate, creatine, and choline, among others.

Functional MRI (fMRI) is performed by exploiting changing regional blood concentrations of oxyhemoglobin and deoxyhemoglobin with the blood oxygen level dependent (BOLD) effect. Oxyhemoglobin is weakly diamagnetic since it has no unpaired electrons, but deoxyhemoglobin is strongly paramagnetic due to four unpaired electrons at each iron center, and the BOLD effect is directly related to the concentration of deoxyhemoglobin in the blood, which causes T2* relaxation. When cortical activation occurs, the T2* shortening effect of deoxyhemoglobin is diminished because there is an influx of oxygenated blood, which is seen as increase in the BOLD signal. This signal is correlated with local field potentials (LFPs), and is modulated by the hemodynamic response function (HRF). Statistical correlations between regions of the brain can be mapped using resting-state fMRI, or with task-based fMRI experiments.

Finally, moving blood can be imaged in a variety of ways using MR angiography. First, time-of-flight (TOF) MRA uses repetitive RF pulses to saturate tissue in a particular region, and then allow fresh spins that have not experienced the saturation pulses to flow into the imaging volume. This yields flow-related enhancement which can be imaged with 2D or 3D T1-W GRE sequences to visualize blood vessels. Spins that move in a magnetic field also accumulate phase, which can be exploited using phase-contrast MRA, in which bipolar gradients intentionally encode a net phase, which can be used to quantify the flow using a specific velocity encoding (VENC) parameter.

This is a small subset of the ways to generate contrast with MRI, but shows the numerous ways that an MR scanner can be used to perform research and aid diagnosis in the clinic.

Acknowledgements

I would like to acknowledge all of the students, residents, and trainees out there. Keep up the hard work!