part iii physics: medical physics magnetic resonance imaging 1999 part iii physics: medical physics...

Part III Physics: Medical Physics Magnetic

Resonance Imaging1999

Part III Physics: Medical Physics Option

Magnetic Resonance Imaging

Dr T A Carpenterhttp://www.wbic.cam.ac.uk/~tac12

Part III Physics: Medical Physics Option Magnetic Resonance Imaging

Lecture Content

Lecture I– Overview of Nuclear Magnetic Resonance

– Excitation and Signal detection– One pulse and Two pulse

experiments– Hardware


Lecture Content

Lecture II– How does NMR become MRI – Effects of Magnetic Field

Gradients– Imaging pulse sequences– contrast– examples


Lecture Content

Lecture III– functional MRI– Diffusion MRI– interventional MRI– examples


Useful Web Sites

Rochester Institute:

http://www.cis.rit.edu/htbooks/mri/mri-main.htm

UCLA Brain Mapping Centre:

http://brainmapping.loni.ucla.edu/BMD_HTML/SharedCode/Shared.htm


NMR History

1921: Compton: electron spin 1924: Pauli: Proposes nuclear spin 1946: Stanford/Harvard group

detect first NMR signal mid -50 to mid 70’s NMR become

powerful tool for structural analysis mid-70 first superconducting

magnets


NMR History

1976: Lauterbur: First NMR image of sample tubes in a chemical spectrometer

1981: First commercial scanners <0.2T

1985: 1.5T scanner 1986: Rapid developments in SNR,

resolution etc 1998: Whole body 8T at OSU


Nuclear Zeeman EffectApplication of strong magnetic field B0 lifts degeneracy of nuclear spin levels

For spin 1/2:

E = h B0

Gyromagnetic ratio (constant of nucleus)

For hydrogen = 42.5 Mhz/T

E


Population Difference

Given by Boltzman Statistics:

nexp( -hBo/kT )n

population difference is small <1 in 106

NMR is very insensitive


Semi-Classical ModelGyroscopic motion of magnetic moment about B0

B0

Use classical mechanics(Larmor)

w0 = - B0


Ensemble Average

M


Rotating FrameConsider precessing moment in a frame of reference rotating at the larmor frequency around B0

xy

= Bo

X’ Y’


Rotating FrameClassical treatment of M

Effect of RF in laboratory Frame:

Y

XEquivalent to sinusoidal Brf


Rotating FrameClassical treatment of M

Effect of RF in rotating Frame:

Y

X X’ Y’

B0

Brf


Signal Detectionrotating Frame:

Y’X’

B0

YX


Fourier Transformation

FT

Sampling frequency = 2 expected frequency spread (Nyquist)

Basic Spin Echo Imaging 28Part III Physics: Medical Physics Option Magnetic Resonance Imaging

Effect of RF pulses:

B1 (rf)

y’ y’

x’ x’

z zB0

90o degree pulse



B1 (rf)

y’ y’

x’ x’

z zB0

90o degre

e pulse



B1 (rf)

y’ y’

x’ x’

z zB0

180o pulse

(inverting pulse)


Effect of 180o RF pulses:

B1 (rf)

y’ y’

x’ x’

z zB0

180o degre

e pulse



B1 (rf)

y’ y’

x’ x’

z zB0

180o degre

e pulse



B1 (rf)

y’ y’

x’ x’



B1 (rf)

y’

x’

y’

x’

y’

x’

y’

x’


Two Pulse sequences (I)

90—— ——90

Saturation recovery

Two Pulse sequences (I)

180—— ——90

Inversion recovery

1 2 3 4 5 6

T1

1 2 3 4 5 6

T1


T1 Spin Lattice Relaxation Time

Describes the return to equilibrium for spins from the excited state

Spins loose heat to the rest of the world

Requires fluctuating magnetic field near the Larmor frequency for an effective transfer of energy from a spin to surrounding lattice


Two Pulse sequences (II)

90—— ——180 —— ——

Spin Echo sequence

y’

x’



90—— ——180 —— ——

Spin Echo sequence

y’

x’

e -t/T2* e -t/T2



90—— ——180 —— ——

Spin Echo sequence

y’

x’


T2 and T2*

e -t/T2* e -t/T2

H

O

H H

O

H


Spin-Spin Relaxation Time

Static inhomogeneities refocussed by 180 pulse

Time varying imhomogeneity are not

T2 changes in disease give rise to diagnostic value of MRI


Superconducting Magnet

Helium vessel containing super-con coil

Vacuum


Superconducting Magnet

Bore B0

100cm 0 4T80cm 0 8T


Shimming


Other Magnet Types

Permanent magnet, e.g. light weight rare earth magnets, <0.3T


Other Magnet Types


Other Magnet Types

Electromagnet <0.3T


Special Superconducting

Magnets Active Shielding

– Extra coils reduce stray field

– Improves siting

10

12

4

2

5mT contour

0.5T wholebody 3T AS wholebody


RF CoilsRemember Brf must be B0

Field is subject, can use solenoid.


RF CoilsRemember Brf must be B0

Field is subject, cannot use solenoid.

Saddle coil, Brf is coil access. Efficiency is low, and homogeneity is poor


How to Make ImagesImpose (separately):

Bz

xBz

yBz

zX gradient

GxY gradient

GyZ gradient

Gz

Typical values are 10-100 mT/m


How to make images

For a Z gradient

-hz +hz

z = -(B0 + Gz.z)


How to make images


Imaging Gradients

Special coils (together with power supplies) provide linear variation in B0 in X, Y and Z directions

Z

B0Z


Imaging Gradients

Special coils (together with power supplies) provide linear variation in B0 in X, Y and Z directions

X,Y


Selection of SliceUse Fourier relationship:

RF Amplitude (volts)


Selection of sliceSlice thickness adjusted by changeimg gradient strength or slice bandwith (longer pulse has narrower frequency spread)

Slice position adjusted by changing the centre frequency of the pulse


k-space

k-space is the raw data space before fourier transformation into the image

2D image will be represented by a 2D array of data points spread throughout k-space

Differing the k-space trajectory will alter image contrast


Image vs k-space

(r) S(k)k(t)=

/2G(t)dt


Image vs k-space

(r) S(k)k(t)=

/2G(t)dt

FT


GE k-space trajectory

(r) S(k)k(t)=

/2G(t)dt

RF

G S

G R

G P

S(t)


(r) S(k)k(t)=

/2G(t)dt

RF

G S

G R

G P

S(t)

-kr +kr



(r) S(k)k(t)=

/2G(t)dt

RF

G S

G R

G P

S(t)

-kr +kr

-kp

+kp



(r) S(k)k(t)=

/2G(t)dt

-kr +kr

-kp

+kp

SE k-space trajectory

RF

GS

GR

GP

S(t)


Definitions

RF

GS

GR

GP

S(t)

TR


Definitions

RF

GS

GR

GP

S(t)

TE


1 2 3 4 5 6

T1

1 2 3 4 5 6

T2

Controlling contrast


1 2 3 4 5 6

T1

1 2 3 4 5 6

T2

Proton DensityTR TE


1 2 3 4 5 6

T1

1 2 3 4 5 6

T2

T2 ContrastTR TE


0.5T Multislic

e Multiech

oTR2000/30..90

30ms 90ms


1 2 3 4 5 6

T1

1 2 3 4 5 6

T2

T1 ContrastTR TE


Effect of Flip angle

X’ Y’

B0

Brf



X’ Y’

B0

Brf

90o pulse

Maximum signal but have to wait 5T1 for recovery



X’ Y’

B0

Brf



X’ Y’

B0

Brf

Flip angle 30o:

detect M0sin = 0.5 M0 remaining M0cos = 0.87 M0


TR/TE/

41/9/15

500/9/15

41/9/9041/9/60

500/9/90

Contrast

versus

Contrast

versus TR


Why ?

freeze involuntary patient motion visualization of dynamic process

– fast imaging: minutes– turbo imaging: seconds

More complex MRI experiments– obtain multiple images vary some

parameter e.g. TI reduce patient examination time


Why does MRI take so long

Answer– Only one phase encode line acquired

per excitation– Spin Echo: 256*3s for T2, 256*0.6s for

T1– Gradient Echo: 256*35ms (but have to

do 3D Solution

– get more phase encode lines per excitation


Echo Planar Imaging

Fastest imaging method Typical AQ time is 30-100ms Low RF deposition Very fast gradient switching Highly demanding on MRI

hardware– B0 homogeneity

– gradient switching


(r) S(k)k(t)=

/2G(t)dt

RF

G S

G R

G P

S(t)

-kr +kr

-kp

+kp

GE-PEI k-space trajectory


(r) S(k)k(t)=

/2G(t)dt

RF

G S

G R

G P

S(t)

-kr +kr

-kp

+kp

GE EPI k-space trajectory


GE vs EP Imaging

TEms

TRms

AQms

BWkhz

GreadmT/m

Switchs

GE 10 35 10 25 2.5 500EPI 50 0.5 250 25 100Assume FOV 25cm AQ = 10ms

Matrix 256 time/sample = 10-2/256

Bandwidth = 25kHz Gread = 25 x 103/0.25

= 100 000Hz/m

= ~ 2.5 mT/m


GE vs EP Imaging

TEms

TRms

AQms

BWkhz

GreadmT/m

Switchs

GE 10 35 10 25 2.5 500EPI 50 0.5 250 25 100Assume FOV 25cm AQ = 0.5ms

Matrix 128 time/sample = 5x10-4/128

Bandwidth = 250kHz Gread = 250 x 103/0.25

= 1 000 000Hz/m

= ~ 25 mT/m


MRI at 3T

128x128 single shot, GE echo planar.

X,Y,Z shim only (~30s)

No template or navigator correction

Straight FFT after row reversal


fMRI (functional MRI)

Monitor T2 or T2* contrast during cognitive task

eg acquire 20-30 slices every 4 seconds

Design experiment to have alternating blocks of task and control condition

Look for statistically significant signal intenisty changes correlated with task blocks


Echo-Planar fMRI

response stimulus

GE-images with EPI fMRI correlation mapsSignal response averaged over region


oxyhaemoglobin

deoxyhaemoglobin

Resting

O2 & glucose


O2 & glucose

Blood flow‘over-compensation’

%O2

Activated

ATP ADP

BOLD signal


Effect of Intravascular Oxygenation level

Blood vessel

Paramagnetic

T2 (and T2*)

reduced because of diffusion through field gradients

Diamagnetic

deoxy oxy


T2* curves activated and rest

time (ms)

signal

activated

rest

TE Signal difference ~ 1-5 %

oxyhaemoglobin

deoxyhaemoglobin

resting activated


Unilateral Finger Opposition (high res)


Definitions

Diffusion relates to the microscopic Brownian thermal motion of molecules

Perfusion, classically is defined as that process that results in the delivery of nutrients to cells, normally expressed as ml/min/100g wet weight of tissue


Effect of Diffusion on NMR

Rms. of an ensemble is zero For a single molecule diffusion results

in a gaussian distribution of displacements

r


Diffusion and Spin echoes


Diffusion and Spin echoes

I/I0 = e -bD

b = 2g22(-/3)


D and ADC

0

2

4

6

8

10

0 500 1000 1500

water

DMSO

I/I0 = e -bD

b = 2g22(-/3)

H2O = 2.1 x 10 -3 mm2s-1

DMSO = 0.55 x 10 -3 mm2s-1

normal = 0.71 x 10 -3 mm2s-1

ischaemic = 0.55 x 10 -3 mm2s-1

b

Log

(I/

I 0)


Diffusion Weighted Imaging

RF

GsGrGp


Diffusion Weighted Imaging

RF

GsGrGp

Gdiffusion


0

2

4

6

8

10

0 500 1000 1500

water

DMSO

Typical Values: = 20, = 50

Gmax b0.5 311 1245 3104

10 12418b

Log

(I/

I 0)


Practical Problems in Human DWI

Gross Motion– Head motion – breathing

Pulsitility– CSF/brain pulsation

Anisotropy– D is direction dependant,

especially white matter


Practical Problems in Human DWI

Gross Motion– Echo Planar Imaging – navigator echoes

Pulsitility– gating plus navigator echoes

Anisotropy– Measure trace, Dxx + Dyy + Dzz– Measure full tensor (all matrix

elements)


Diffusion Weighted EPI (b=1570 s/mm2)

READ PHASE SLICE

FOV 25cm, TE 118ms TYDW-EPI 128x128 interpolated to 256x256Partial k-acquisition (62.5%)4 interleaves, = 28ms ; = 66 ms


Cambridge NIH van Zijl

AD

C t

race




An

istr

op

y In

dex


MRI and O15 water PET


Gadolinium blous experiment in rat brain

Image number (relative to blous injection)

-20 -10 0 10 20 30 40 50 60

Rel

axat

ion

rate

cha

nge

(s-1

)

-1

0

1

2

3

4

5

6


Effect of Intravascular Gd

Blood vessel

Tissue

Tissue


Effect of Intravascular Gd

Blood vessel

Tissue

Tissue

T2 (and T2*)

reduced because of difussion through field gradients


Gadolinium blous experiment in rat brain

Image number (relative to blous injection)

-20 -10 0 10 20 30 40 50 60

Rel

axat

ion

rate

cha

nge

(s-1

)

-1

0

1

2

3

4

5

6


Data Analysis

Fit first pass of the bolus (avoid recirculation)

Gamma variate, or (better) Monte Carlo

Estimate arterial input function from large vessel signal

rrCBV, rrCBF but absolute MTT


Perfusion weighted MRI of a patient with a high grade stenosis (>90%) of the right internal carotid artery leading to a terminal supply zone infarction in the region of the middle cerebral artery, from http://www.picker.com/mr/acr/perfusn/perfusn.htm

T2 weighted FSE images (3555/80/4)

rrCBV-map map of the bolus delay (MTT image)


Caution

Numbers obtained are not for true perfusion (as measured by PET)

Similar to dynamic CT, DSC measures micro-capillary flow

However good correlation between PET and DSC (in pigs), in humans??


True Perfusion by MRI

Arterial spin labeling– EPISTAR, ASL, QUIPS– label arterial blood on the way

into brain– subtract images with and without

labelling– difference is due to arterial water

that has entered tissue, i.e. perfusion


Scanner Overview

part iii physics: medical physics magnetic resonance imaging 1999 part iii physics: medical physics...

Documents

perfusion slide

bolus delay mtt image

nmr signal mid

carpenter http

nuclear spin

nmr image of sample

arterial water

electron spin