Top Banner
The Basics of MRI Ives Levesque BIC Seminar Series, October 5th, 2009 McConnell Brain Imaging Centre Montreal Neurological Institute McGill University NOT FOR RE-USE
68

The Basics of MRI - McConnell Brain Imaging Centre › ...10-05_-_The_Basics_of_MRI_-_Ives_Levesq… · The Basics of MRI Ives Levesque BIC Seminar Series, October 5th, 2009 McConnell

Jun 08, 2020

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
  • The Basics of MRIIves Levesque

    BIC Seminar Series, October 5th, 2009

    McConnell Brain Imaging CentreMontreal Neurological InstituteMcGill University

    NOT F

    ORRE

    -USE

  • Acknowledgments for material

    Prof. Bruce PikeIlana LeppertDr. Jennifer CampbellMichael FerreiraChristine TardifDr. Luis Concha

    NOT F

    ORRE

    -USE

  • MRI: a simple overview

    somethinghappens

    put subjectin scanner

    knock!buzz! knoc

    k!

    bang!

    NOT F

    ORRE

    -USE

  • MRI: behind the scenes

    Fourier transform

    B0

    NMRGx

    z

    y

    encoding

    x

    z

    y

    M

    B1

    excitation

    detection

    NOT F

    ORRE

    -USE

  • Outline

    • NMR physics• Spatial encoding and reconstruction• Basic MRI sequences: GE, SE• Special contrast: DWI, BOLD• Trade-offs and limitations• Image artefacts• SafetyN

    OT FO

    RRE

    -USE

  • NMR Physics

    NOT F

    ORRE

    -USE

  • NMR: Nuclear spin• “odd-numbered” nuclei have net angular

    momentum• magnetic moment, or spin, designated by

    vector µ

    • In biological tissue: 1H, 23Na, and 31P

    N

    SNOT F

    ORRE

    -USE

  • NMR: spins in magnetic field• magnetic moments (spins) align to external magnetic• net excess of spins in parallel state (lower energy)

    state of net magnetization, vector M

    B0

    external field B0

    M

    ∑=i

    inet µMNO

    T FOR

    RE-U

    SE

  • Spins in a magnetic field experience a torque:

    Static field produces precession about at Larmor frequency

    γ/2π = 42.58 MHz/T for 1H, f ~ 64 MHz at 1.5 T

    This is the resonance frequency

    NMR - Free precession & resonance

    0Bf γ=

    BMM γ×=dt

    dB0

    NOT F

    ORRE

    -USE

  • • Superconducting solenoid magnet

    • Liquid helium cooling

    • Many km’s of wire• 1.0 – 9.4 tesla

    The main magnet

    B0NOT

    FOR

    RE-U

    SE

  • Manipulating magnetization

    • Apply magnetic field B1– Perpendicular to B0– Amplitude ~ 20µT– Alternating with frequency f1

    • Rotate M from z-axis to transverse (xy) plane

    – Most efficient on resonance (f1 = f0)

    y

    xz

    NOT F

    ORRE

    -USE

  • Excitation (stationary frame of reference)

    B1 field in yellow, magnetization vector in red

    NOT F

    ORRE

    -USE

  • The rotating frame of reference

    f0 = γ B0

    y

    x

    z

    B0y’

    x’

    z

    M M

    Mxy

    Mz

    • “laboratory” frame of reference

    • vector M precesses at resonant frequency

    • rotating frame of reference (at f0)

    • “ignore” precession of M

    NOT F

    ORRE

    -USE

  • Excitation (rotating frame of reference)

    B1 field in yellow, magnetization vector in red

    NOT F

    ORRE

    -USE

  • Relaxation

    • M returns to equilibrium

    • Recovery of Mz• Decay of Mxy• Each appens at

    a different rate

    y’

    x’

    z

    B0 Recovery of Mz

    Decay of Mxy

    *rotating frame of reference!

    NOT F

    ORRE

    -USE

  • T1 Relaxation: Recovery of Mz

    • Energy exchange with surrounding medium

    0 1 2 3 40

    0.2

    0.4

    0.6

    0.8

    1Mz recovery: T1

    ρ(1- e-t/T1)

    Mz

    y’

    x’

    z

    B0 T1 relaxation

    Time (s)

    NOT F

    ORRE

    -USE

  • T2 Relaxation: Decay of Mxy

    • Effect of spin-spin interaction

    0 0.1 0.2 0.3 0.40

    0.2

    0.4

    0.6

    0.8

    1Mxy decay: T2

    ρe-t/T2

    Mxy

    y’

    x’

    z

    B0

    T2 relaxationTime (s)

    NOT F

    ORRE

    -USE

  • T2* Relaxation: Faster decay of Mxy

    • Additional effect of B0 inhomogeneity

    • Reversible

    y’

    x’

    z

    B0

    T2* relaxation > T2 relaxation

    0 0.1 0.2 0.3 0.40

    0.2

    0.4

    0.6

    0.8

    1

    Time (s)

    Mxy decay: T2, T2*

    Mxy

    ρe-t/T2*

    NOT F

    ORRE

    -USE

  • Relaxation in the lab frame

    Individual spins in blue, Mxy in yellow, Mz in red

    NOT F

    ORRE

    -USE

  • Image contrast• Signal depends on:

    – sequence parameters (TR, TE, α)– NMR parameters in tissue: PD, T1, T2, T2*– diffusion, MT, iron, contrast agents, etc.

    • Each pulse sequence has its own behavior

    T1W T2W PD or “Intermediate”

    NOT F

    ORRE

    -USE

  • T1 & T2 relaxation in CNS at 3 T

    0 0.6 0.80

    0.1

    0.2

    0.3

    0.4

    0.5

    0.6

    0.7

    0.8

    0.9

    1

    Time (s)

    T2 relaxation

    0 1 2 3 40

    0.1

    0.2

    0.3

    0.4

    0.5

    0.6

    0.7

    0.8

    0.9

    1

    Time (s)

    T1 relaxation

    WM

    GM

    GM

    WM

    CSF

    CSF

    0.2 0.4

    Mz Mxy

    GM: T1 = 1300 ms T2 = 100 ms ρ = 0.8WM: T1 = 800 ms T2 = 80 ms ρ = 0.7

    CSF: T1 = 5500 ms T2 = 2200 ms ρ = 1.0NO

    T FOR

    RE-U

    SE

  • NMR signal detection

    f0x

    y

    z

    B0

    stationary (lab) frame of reference0 0.2 0.4 0.6 0.8 1

    -1

    0

    1

    Time (ms)

    Sig

    nal

    “FID”

    f0

    T2* decay

    NMR signal received in RF coil (Faraday’s Law)

    NOT F

    ORRE

    -USE

  • • excitation– apply magnetic field B1, perpendicular to B0 , at

    resonant frequency f0, to rotate M0 into x-y plane• evolution

    – M0 precesses freely, relaxing to equilibrium on z-axis– energy released in relaxation process at frequency f0

    • detection– detect signal: changing magnetic flux in x-y plane

    produces a voltage in receive coil

    • waiting (for TE) and spatial encoding

    • repetition (after TR)

    Beyond exploiting the resonance

    NOT F

    ORRE

    -USE

  • Spatial Encoding and Reconstruction

    NOT F

    ORRE

    -USE

  • Magnetic field gradients

    • Linear variations in magnetic field (millitesla / meter)

    • Manipulate frequency of precession• Encode spatial information in signal

    – frequency (readout)– phase (prior to readout)NO

    T FOR

    RE-U

    SE

  • Gradients modify the amplitude of B0

    B0

    Gz= dB0/dz

    B0 Gy= dB0/dy

    NOT F

    ORRE

    -USE

  • B0

    frequency (Hz)

    MRI: localization in 1-D

    Paul Lauterbur

    FT

    f0time (ms)

    sign

    al

    f0 = γ B0Gx f > f0f < f0

    N samples N points

    NOT F

    ORRE

    -USE

  • kx

    ky

    x

    y

    2D FT

    magnitude k-space data magnitude reconstructed image

    2D-FT imaging

    N1 samples per TR

    N2 repetitions of TR N2 voxels

    N1 voxels

    NOT F

    ORRE

    -USE

  • 3D-FT imaging

    3D FT

    NOT F

    ORRE

    -USE

  • 3D-FT imaging

    NOT F

    ORRE

    -USE

  • Echo-planar imaging (EPI)

    • Acquire entire 2D space in one shotky

    kx

    RF

    Gy

    Gx NOT F

    ORRE

    -USE

  • Example of alternate k-space trajectory

    • interleaved spirals

    kx

    ky

    RF

    Gx

    Gy NOT F

    ORRE

    -USE

  • Basic MRI sequences

    NOT F

    ORRE

    -USE

  • Breakdown of MRI experiment: the “sequence”

    acquisition

    TETR

    preparation

    - inversion preparation- MT saturation

    relaxation

    - acquire other slices- spoil or rewind

    excitation

    - RF pulses

    - refocusing pulses - selective or non-selective

    encoding

    - gradients

    - diffusion encoding

    acquisition

    - gradients- signal detectionNO

    T FOR

    RE-U

    SE

  • The gradient echo sequence

    acquisition

    TETR

    preparation

    - inversion preparation (MP-RAGE)

    - MT saturation

    relaxation

    2D: acquire other slices

    3D: spoil or rewind

    excitation

    - excitation pulse

    2D: slice-selective3D: non- or slab-selective

    encoding

    - phase encoding gradients

    acquisition

    - read-out gradients- data acquisition

    - single-line or EPI

    NOT F

    ORRE

    -USE

  • GE images depend on T1 and T2*

    T1w FLASH

    Localizer (“scout”) Post-gadolinium T1w

    GE EPINO

    T FOR

    RE-U

    SE

  • Spin echo

    • T2* decay caused by spin dephasing

    • Any pair of RF pulses can reverse (or refocus) T2’ to some extent

    • 90°-180° combination is most effective• SE signal decays with time T2

    22*

    2

    111TTT ′

    +=

    NOT F

    ORRE

    -USE

  • The spin echo sequence

    acquisition

    TRTE/2TE/2

    relaxation

    2D: acquire other slices

    excitation

    - excitation pulse- refocusing pulse

    2D: slice-selective

    encoding

    - phase encoding gradients

    read-out

    - read-out gradients- data acquisition

    - single-line or “turbo”

    NOT F

    ORRE

    -USE

  • NOT F

    ORRE

    -USE

  • Turbo spin echo

    acq. acq. acq.

    2D FT

    N2 / 3 repetitions of TR N2 voxels

    3 × N1 samples per TR N1 voxels

    NOT F

    ORRE

    -USE

  • Turbo spin echo images

    TR/TE = 2.7s/12ms, ETL 3, 256x256, 60 slices

    TR/TE = 5.4s/83ms, ETL 7, 256x256, 60 slicesNO

    T FOR

    RE-U

    SE

  • Diffusion contrast in MRI

    NOT F

    ORRE

    -USE

  • Gradients and phaseGy

    timeB0 B0 + Gy

    Gy

    B0

    Total:Total:

    NOT F

    ORRE

    -USE

  • Gradient dephasing and rephasingG

    time

    B0

    Total:

    B0 - G B0

    Total:

    B0 + G

    NOT F

    ORRE

    -USE

  • Gradients, phase, and motionG

    time

    B0B0 + G B0 B0 - G

    NOT F

    ORRE

    -USE

  • Diffusion MRI SE sequence

    EPI

    diffusion encoding gradients

    ×NO

    T FOR

    RE-U

    SE

  • Diffusion indices

    FA: anisotropy

    index

    RGB plot: principal direction,

    scaled by FA

    trace of diffusion tensor ~ Dav

    NOT F

    ORRE

    -USE

  • BOLD MRI

    NOT F

    ORRE

    -USE

  • O2

    at rest

    O2

    ACTIVATED

    BOLD contrast

    ∆T2*

    NOT F

    ORRE

    -USE

  • Hemoglobin and T2*

    deoxy-Hbχ ~ 1.6

    oxy-Hbχ ~ -0.3

    longer T2, T2*

    ↑ BOLD upon↑ activation

    shorter T2,T2*

    baseline BOLD

    oxy-RBCdeoxy-RBC

    B0

    NOT F

    ORRE

    -USE

  • fMRI acquisition

    • series of stimulus conditions–activation and baseline

    off

    on

    stimulus

    rapid image acquisition NO

    T FOR

    RE-U

    SE

  • fMRI example: finger tapping task

    NOT F

    ORRE

    -USE

  • GE-BOLD vs SE-BOLD

    Vessel Radius [µm]

    ∆R2*

    [s-1

    ]

    GE BOLD (TE = 40ms)

    Vessel Radius [µm]

    ∆R2

    [s-1

    ]

    SE BOLD (TE = 100ms)

    • GE-BOLD offers greater sensitivity• SE-BOLD more specific to small vessels

    Buxton, 2002NO

    T FOR

    RE-U

    SE

  • Choices, trade-offs and limitations

    NOT F

    ORRE

    -USE

  • Voxel-size, FoV, and scan time are linked

    N2 voxelsN2 repetitions of TR

    2D FT

    N1 samples per TR N1 voxels

    scan duration = TR × N2 × NslicesNO

    T FOR

    RE-U

    SE

  • Voxel size & scan time: link to SNR

    ( )21,, TTfTVSSNR readvoxel

    S

    ρσ

    ∝≡

    Basic T1w

    1×1×1 mm3

    256×256×160

    1

    1

    15.7 minutes

    Voxel size

    Matrix size

    # averages

    rel. SNR

    Acq. time

    High-res

    0.5×0.5×1 mm3

    512×512×160

    1

    0.25

    31.4 minutes

    Thick slice

    1×1×2 mm3

    256×256×80

    1

    2

    7.9 minutes

    High-res + SNR

    0.5×0.5×1 mm3

    512×512×160

    16

    1

    8.4 hours!

    NOT F

    ORRE

    -USE

  • + Better SNR (up to 2×)+ Higher BOLD sensitivity

    • Longer T1

    - Higher energy deposition- B1 less homogeneous- More B0 artefacts

    Impact of field strength

    MP-RAGE, C. Tardif

    1.5 T 3 T

    NOT F

    ORRE

    -USE

  • Artefacts:what’s wrong with my image?

    NOT F

    ORRE

    -USE

  • Subject motion affects images

    phase encoding

    • Motion during scan corrupts phase encoding• Solutions: subject co-operation, immobilization, fast

    imagingNO

    T FOR

    RE-U

    SE

  • Flow artefacts

    2D SETR 3s

    TE 50ms

    3D slab-selective FLASHNO

    T FOR

    RE-U

    SE

  • Ghost sightings in EPI imagesphase encoding

    signal × 10• Imperfect gradient behaviour results in ghosting• Solutions: select phase encode direction to avoid

    NOT F

    ORRE

    -USE

  • Magnetic susceptibility artefacts

    • Changes in susceptibility affect B0 homogeneity– Image distortion and signal pile-up

    • Solutions: parallel imaging, shorter TE, angulation

    GR

    EE

    PI

    NOT F

    ORRE

    -USE

  • MRI Safety

    NOT F

    ORRE

    -USE

  • MRI safety: the main magnetic fieldThe magnet is always on, and will “perturb” instruments

    NOT F

    ORRE

    -USE

  • MRI safety: Peripheral nerve stimulation• Magnetic field variations in time (dB/dt) induce

    currents– Gradient coils switch on/off in less than 1 millisecond

    • Currents induced in the subject can produce nerve stimulus

    • Effect is greatest at the ends of the coils, because gradient amplitude is greater there

    B0

    Gz=dB0/dz

    NOT F

    ORRE

    -USE

  • MRI safety: Energy deposition

    • B1 field deposits energy into the body, results in heating

    • SAR = specific absorption rate– power per unit (subject) mass, W/kg

    • Heat is dissipated by the body’s own cooling, with the help of ventilation

    • Monitored by the scanner

    TRPSAR RFRFτ∝

    NOT F

    ORRE

    -USE

  • Summary

    Fourier transform

    B0

    NMR Gy

    z

    x

    encoding

    y

    z

    x

    M

    B1

    excitation

    detection

    NOT F

    ORRE

    -USE

  • Useful referencesMRI From Proton to Picture

    DW McRobbie, EA Moore, MJ Graves, Cambridge Press, 2002 (preview on Googlebooks)

    Introduction to Functional Magnetic Resonance Imaging: Principles and TechniquesRB Buxton, Cambridge Press, 2002 (preview on Googlebooks)

    Principles of magnetic resonance imagingDwight Nishimura, 1996 (out-of-print )

    Magnetic Resonance Imaging: Physical Principles and Sequence DesignEM Haacke, Wiley, 1999

    FMRIB MRI physics lectureshttp://www.fmrib.ox.ac.uk/education/graduate-training-course/program/lectures/mri-physics/mri-physics-course

    The Basics of MRI (Joseph Hornak)http://www.cis.rit.edu/htbooks/mri/

    NOT F

    ORRE

    -USE

    The Basics of MRIAcknowledgments for materialMRI: a simple overviewMRI: behind the scenesOutlineNMR PhysicsNMR: Nuclear spinNMR: spins in magnetic fieldNMR - Free precession & resonanceThe main magnetManipulating magnetizationExcitation (stationary frame of reference)The rotating frame of referenceExcitation (rotating frame of reference)RelaxationT1 Relaxation: Recovery of MzT2 Relaxation: Decay of MxyT2* Relaxation: Faster decay of MxyRelaxation in the lab frameImage contrastT1 & T2 relaxation in CNS at 3 TNMR signal detectionSpatial Encoding and ReconstructionMagnetic field gradientsGradients modify the amplitude of B0Basic MRI sequencesBreakdown of MRI experiment: the “sequence”The gradient echo sequenceGE images depend on T1 and T2*Spin echoThe spin echo sequenceTurbo spin echoTurbo spin echo imagesDiffusion contrast in MRIDiffusion MRI SE sequenceDiffusion indicesBOLD MRIBOLD contrastHemoglobin and T2*Choices, trade-offs and limitationsVoxel-size, FoV, and scan time are linkedVoxel size & scan time: link to SNRArtefacts:what’s wrong with my image?Subject motion affects imagesFlow artefactsGhost sightings in EPI imagesMagnetic susceptibility artefactsMRI SafetyMRI safety: the main magnetic fieldMRI safety: Peripheral nerve stimulationMRI safety: Energy depositionSummaryUseful references