Magnetic Resonance Imaging: Advanced Image Acquisition Methods, Artifacts, Spectroscopy, Quality Control, Siting, Bioeffects, and Safety



Magnetic Resonance Imaging: Advanced Image Acquisition Methods, Artifacts, Spectroscopy, Quality Control, Siting, Bioeffects, and Safety





13.0 INTRODUCTION

Advanced pulse sequences and fast image acquisition methods; methods for perfusion, diffusion, and angiography imaging; spectroscopy; image quality metrics; common artifacts; MR siting; and MR safety issues are described and discussed with respect to the underlying physics.


13.1 IMAGE ACQUISITION TIME



  • 1. Acquisition time, 2D acquisition



    • a. Time = TR × #PEG × NEX—where TR is the repetition time, #PEG is the number of phase-encode gradient applications, and NEX is the number of excitations (averages).


    • b. Matrix size defining k-space is often not square—typically, the smaller dimension is assigned to PEG.


    • c. Tradeoff of time and SNR are considered.


  • 2. Multislice data acquisition



    • a. During the TR, cycling gradients and tuning RF excitation frequency images the volume (Fig. 13-1).


    • b. Tradeoff is cross excitation of adjacent tissues and loss of contrast from nonsquare excitation pulses.


    • c. Total number of slices = TR/(TE + C), where C is a constant dependent on MR equipment capabilities.


    • d. Longer TR (e.g., T2-weighted SE) can have more slices acquired in the acquisition.







  • 3. Acquisition time, 3D acquisition



    • a. The 3D acquisition is initiated with a slice encode excitation in addition to a phase-encode excitation.


    • b. Time (3D) = TR × # phase-encode steps (Z-axis) × phase-encode steps (X-axis) × NEX.



13.2 FAST IMAGING TECHNIQUES



  • 1. Fast pulse sequences



    • a. Fast spin echo (FSE) uses multiple PEG steps with multiple 180° refocusing pulses per TR.



      • (i) Multiple lines in k-space are filled per TR resulting in an echo train length (ETL) (Fig. 13-2).


      • (ii) Speed increase is acquisition time × 1/ETL.


      • (iii) Characteristics: high SAR, good immunity from inhomogeneities (with 180° excitations).







    • b. Echo planar image (EPI) acquisition uses oscillating readout gradient and phase-encode gradient “blips.”



      • (i) Can be initiated with spin echo or gradient echo (spin echo EPI—Fig. 13-3)


      • (ii) Offers “snapshot” capability: down to 50 ms acquisition


      • (iii) Characteristics: low resolution, artifacts (ghosting and geometric distortion due to susceptibility)







    • c. GRASE (gradient and spin echo) sequence combines FSE and EPI.



      • (i) A series of GREs between inversion RF pulses is repeated over multiple fast spin echoes.


      • (ii) Achieves benefits of GRE (speed) and SE (RF refocusing to compensate for T2* effects).


      • (iii) For details, see textbook Page 499, Figure 13-4.


  • 2. k-Space filling



    • a. Nonsequential filling of k-space data to obtain optimal contrast (Fig. 13-4).


    • b. Centric k-space filling for SNR advantage when echoes have their highest amplitude.


    • c. Keyhole filling collects central lines later for important events such as contrast-enhanced angiography.








  • 3. Noncartesian k-space acquisition (nonrectilinear filling—Fig. 13-5)



    • a. Radial imaging—generates radial spokes passing through the center of k-space in 2D and 3D space



      • (i) Yields higher sampling density at the center than the periphery


      • (ii) Benefits applications of dynamic imaging requiring high temporal resolution (e.g., contrast-enhanced angiography, cardiac imaging; see text book, Fig. 13-7 for details)







    • b. Spiral imaging—an alternate method of EPI, involving oscillation of equivalent encoding gradients


    • c. Propeller—acquisition technique that mitigates motion artifacts



      • (i) Rectangular block of data (“a blade”) is acquired and rotated about center of k-space.


      • (ii) Redundant information is used to identify and correct motion artifacts (see textbook Pg. 502-503, Fig. 13-8 for details).


  • 4. Data synthesis—takes advantage of the symmetry and redundancy of k-space frequency domain signals



    • a. Fractional NEX: in phase-encode direction, the number of excitations are reduced (Fig. 13-6, left).


    • b. Fractional echo: in frequency-encode direction, a fraction of echo reduces TE, and more slices can be acquired in one TR period (Fig. 13-6, right).


    • c. Tradeoff for both methods: loss of SNR due to fewer excitations per voxel in the volume.








  • 5. Parallel imaging



    • a. Response of multiple-receive RF coils overcome aliasing artifacts due to undersampling.


    • b. SENSitivity Encoding (SENSE) uses the sensitivity profile of each coil element.


    • c. A k-space-based approach synthesizes skipped lines directly (see textbook, Pg. 505, Fig. 13-10).



      • (i) Method: GeneRalized Autocalibrating Partially Parallel Acquisition (GRAPPA)


    • d. Scan time reduction factor can be higher than two, but SNR is reduced.


    • e. SNR is inversely proportional to the coil geometry factor and square root of reduction factor:

      image, where R is the scan time reduction and g is the coil geometry factor.


  • 6. Multi-band imaging (MB)



    • a. MB further accelerates image acquisition by exciting multiple slices simultaneously (Fig. 13-7).


    • b. No SNR penalty from scan time reduction is incurred as multiple excitation pulses are acquired.


    • c. Coil g-factor in slice direction reduces SNR slightly.







13.3 SIGNAL FROM FLOW



  • 1. Flow-related enhancement



    • a. A process causing increased signal enhancement of moving tissue (blood, CSF).


    • b. High intensity is caused by wash-in of unsaturated protons into a partially saturated volume.


    • c. Elimination of bright signals can be achieved with saturation pulses outside of the imaging volume.


  • 2. MR angiography—exploitation of blood flow enhancement



    • a. Time-of-flight angiography



      • (i) Relies on flow enhancement of “tagged” or “unsaturated” protons into the imaging volume.


      • (ii) Longitudinal magnetization differences of moving blood results in differential vessel contrast.


      • (iii) Use of poor anatomic contrast imaging (e.g., GRASS-FISP sequence) allows use of maximum intensity projections to generate angle-specific views of the vasculature (Fig. 13-8).








    • b. Phase-contrast angiography



      • (i) Relies on the phase change occurring in moving protons (Fig. 13-9).


      • (ii) The phase change is dependent on bipolar gradients in two excitations with opposite polarity.


      • (iii) Time, ΔT, between bipolar gradients is the velocity encoding (VENC) time to ensure optimal phase shift for measurements without aliasing and for gray scale encoding of velocity.


      • (iv) Measurements are quantitative for velocity and direction (Fig. 13-10).












  • 3. Gradient moment nulling (for flow compensation)



    • a. Flowing blood often causes flow artifacts due to the phase dispersion of moving spins (Fig. 13-11).


    • b. Additional gradients set the phase evolution of stationary and moving spins to 0 prior to data collection.








13.4 PERFUSION AND DIFFUSION CONTRAST IMAGING

Perfusion is the delivery of blood to a capillary bed in tissue and permits the delivery of oxygen and nutrients to the cells and removal of waste (e.g., CO2) from the cells—methods include ASL, DSC, DCE (Table 13-1).










  • 1. Arterial spin labeling (ASL)



    • a. ASL uses blood magnetization and measures blood flow (Fig. 13-12).


    • b. Pulsed ASL tags blood with inversion or saturation pulse.



      • (i) Tagged blood moves into region of interest with imaging initiated.


      • (ii) After waiting time (1 to 2 s), another image set without inversion is acquired in same region.


      • (iii) Subtraction of the images removes the signals from static tissues.


    • c. Continuous ASL uses a long RF pulse applied in a plane while the blood signal is inverted.



      • (i) After “post-labeling delay” images in the target area are acquired.


      • (ii) Another acquisition of noninverted spins and subtraction yields the signals from the blood.


    • d. Method suffers from low SNR, requiring multiple signal averages (20 to 60).


    • e. Most important is choice of waiting time—a tradeoff of measuring vascular signals versus tissue perfusion and the T1 decay of blood signals.








  • 2. Dynamic susceptibility contrast (DSC)



    • a. Use of contrast material tagged with susceptibility agents such as gadolinium (Gd).


    • b. T2 and T2* parameters generate large signal differences in the vasculature—Gd concentrations can be tracked as a function of time in arteries and tissues (Fig. 13-13).


    • c. Blood flow can be estimated by deconvolution of the tissue residue function (Fig. 13-14).



      • (i) Mean transit time (MTT) is estimated with ratio of cerebral blood volume (CBV) and cerebral blood flow (CBF) or an integral of blood flow.


    • d. Concern with DSC is contrast agent leakage to extravascular or extracellular space, as the conventional DSC model is based on no leakage.












  • 3. Dynamic contrast enhanced (DCE)



    • a. Purpose is the measure the amount of leakage into tissues.


    • b. “Tofts” and “Extended Tofts” models are used (see textbook Pg. 515-516 for details).


    • c. Dynamic T1-weighted sequence such as SPGR is used.


    • d. Mapping of T1 signal change when Gd contrast agent is injected (Fig. 13-15).







  • 4. Diffusion MRI

    Molecular diffusion is the stochastic translational motion of molecules also known as Brownian motion. Diffusion MRI sequences use strong MR gradients applied symmetrically about the refocusing pulse to produce signal differences based on the mobility and directionality of water diffusion.




    • a. Diffusion-weighted imaging (DWI)



      • (i) Symmetrical gradients of amplitude G and duration δ placed before/after the 180° pulse (Fig. 13-16).


      • (ii) Tissues with more mobility are dephased by the gradients and have a smaller signal than those with restricted diffusion (e.g., ischemic injury).


      • (iii) A T2-weighted image without diffusivity weighting is compared to a weighted image (b value).


      • (iv) A b value (s/mm2) is a diffusion sensitivity factor, and the diffusivity D is the diffusion rate (mm2/s).

        Mxy (b,TE) = M0e-TE/T2ebD


      • (v) The b values are typically in the range of 200 to 2,000.


      • (vi) A higher b value (e.g., b = 1,000) generates more sensitive but noisier diffusion-weighted images.


      • (vii) Apparent diffusion coefficient (ADC) images are generated from the image pair (Fig. 13-17).












    • b. Diffusion tensor imaging



      • (i) Advanced form of diffusion imaging that uses encoding directionality to indicate the anisotropy of white matter by measuring the diffusion restriction and providing structure of surrounding tissues.


      • (ii) A diffusion tensor provides quantitative diffusion metrics (diffusivity and fractional anisotropy).


      • (iii) A minimum of six diffusion-encoding directions are required to generate the tensor; many more are actually used to improve the SNR.


      • (iv) DTI images are color encoded to illustrate directions of diffusion (Fig. 13-18).








13.5 OTHER ADVANCED TECHNIQUES



  • 1. Functional MRI



    • a. Technique of mapping neuronal activities in the brain and relying on a local reduction of deoxyhemoglobin, which is paramagnetic, while the neurons in the region become active


    • b. Blood oxygen level-dependent (BOLD) acquisition



      • (i) A series of dynamic T2*-weighted images using fast EPI GRE sequences map the brain’s active regions through correlation of signals with a repetitive task (e.g., physical, sensory, or cognitive).


      • (ii) Resultant areas are color mapped onto the gray scale images (see textbook, Pg. 521, Fig. 13-26).


    • c. Combination diffusion MRI and fMRI can help visualize structure and function (e.g., identify areas prior to resection of a brain tumor) (see textbook, Pg. 521, Fig. 13-27).


  • 2. Susceptibility-weighted imaging



    • a. Based on T2*-weighted magnitude and phase images obtained with a 3D GRE sequence.


    • b. Images are sensitive to local susceptibility differences from deoxyhemoglobin in venous blood, methemoglobin in blood hemorrhage, and iron deposition.


    • c. Technique utilizes small and local phase differences and magnitude change with image processing to remove phase changes due to field inhomogeneities.


    • d. SWI images are generated from a multiplication of phase masks with the magnitude images to emphasize small phase deviations caused by the susceptibility agents (Fig. 13-19).


    • e. Improved visualization is obtained with minimum intensity projection (mIP) of the SWI image stack.







  • 3. MR elastography



    • a. Technique evaluates the stiffness of tissues with the use of an external mechanical wave generator.


    • b. GRE imaging with motion-encoding gradients (similar to phase-contrast MRA) are used for acquisition.



      • (i) 4 to 8 dynamic images are acquired of magnitude and phase during mechanical vibration.


      • (ii) Displacement information is extracted from the dynamic phase information at the applied frequency.


      • (iii) Submicrometer motion in the direction of mechanical wave is inversely proportional to the wave speed, from which the stiffness is determined, as stiffer tissues have faster wave speed.


    • c. Typical use is assessment of tissue stiffness in the liver (Fig. 13-20).







  • 4. Magnetization transfer contrast



    • a. The interaction between protons in free water molecules and protons hydrated to the macromolecules of a protein provides a conduit for partial saturation with off-resonance pulse.


    • b. Magnetization exchange occurs between the two proton groups.



    • c. Selective saturation of the protons in the hydration layer occurs separately from the bulk water by using narrowband RF pulses tuned to the hydration layer with an off-resonance peak of approximately 5 kHz.


    • d. Transfer of the magnetization partially saturates the adjacent protons in bulk water.


    • e. Saturation of these spins reduces anatomic contrast and is often used in conjunction with MRA time-of-flight methods and other examinations where selective reduction of high signal assists in diagnosis, such as knee cartilage (see textbook, Pg. 522-523, Fig. 13-30).


  • 5. Magnetic resonance spectroscopy (MRS)



    • a. A method to evaluate tissue chemistry by recording and measuring signals from metabolites.


    • b. MRS metabolic peaks are separable by frequency shifts relative to the frequency of water proton.


    • c. Metabolite precessional frequencies are bounded by that of water and fat requiring selective saturation pulses to allow the quantitative evaluation of the very weak metabolite signals (Fig. 13-21).



      • (i) CHESS (Chemical Shift-Selective) or STIR (Chapter 12) signal suppression methods are used.


    • d. MRS signals are derived from proton metabolites in targeted tissues (Fig. 13-22, left, and Table 13-2).



      • (i) Chemical shifts occur due to electron cloud shielding—the amount of shift identifies the metabolite.


    • e. Localization of the target volume is achieved with single voxel or multivoxel techniques.



      • (i) Single voxel sampling areas are about 1 cm3—a STEAM (STimulated Echo Acquisition Mode) or a PRESS (Point REsolved SpectroScopy) sequence are employed to identify the volume.


      • (ii) Multivoxel MRS uses a CSI (chemical shift imaging) technique to delineate multiple voxels of approximately 1 cm3 in 1, 2, or 3 planes over a block of several centimeters of tissue.


    • f. Magnetic resonance spectroscopic imaging (MRSI) generates the signal intensity of a single metabolite in each voxel and color encodes the values (e.g., the choline/creatine ratio), which are superimposed as a map on the anatomical MR image (Fig. 13-22, right) indicating normal/abnormal tissues (Table 13-3).