Basics of MRI

Professor Sir Michael Brady FRS FREng Department of Engineering Science Oxford University Michaelmas 2004

Lecture 1: MRI image formation

Basic nuclear magnetic resonance NMR NMR spectroscopy Magnetic field gradients spatial resolution Extension to 2D, 3D images and volumes

T1 Weighted Image
MR image of a horizontal slice through the brain. In this T1-weighted image, grey matter is lightly coloured, while white matter appears darker.

T1/s white matter grey matter CSF 0.7 1 4 1.5T

R1/s-1 1.43 1 0.25

SPGR, TR=14ms, TE=5ms, flip=20

Brain image analysis

Accurate shape modelling and measurement

T2 Weighted Image

T2/ms CSF grey matter white matter 1.5T 500 8090 7080

SE, TR=4000ms, TE=100ms

Over the past 20 years, we have developed new ways to image anatomy, new ways to see inside the body, noninvasively

We can watch the body in action, as it responds to the injection of a drug or contrast agent, to highlight aberrant physiology

We can watch the body functioning in a whole range of ways the brain thinking, degradation in white matter, and the pulsing of the heart

Now we are beginning to image cellular and molecular processes the convergence of molecular biology and image analysis

MRI machines
Magnet

bed

Siemens Avanto 1.5T state-of-theart MRI machine

The Magnet
Magnets field strength: imaging 0.2T to 2.0T spectroscopy- 2.0T to 7.0T Low field Intermediate High field Ultra high 0.2- 0.5T 0.5- 1.5T 1.5- 4.0T > 4.0T Earths magnetic field = 510-5 Tesla

MRI System Block Diagram

RF amp X Y Z amp amp amp

spectrometer

r.f.coil

gradient coil

MRI is considered ideally suited for soft tissue problems

*** MRI is to soft tissue as x-ray is to dense tissue (bone)***

Diagnosing multiple sclerosis (MS) Diagnosing brain tumours Diagnosing spinal infections Visualizing torn ligaments in the wrist, knee and ankle Visualizing shoulder injuries Evaluating bone tumours, and herniated discs in the spine Diagnosing strokes in their earliest stages

Some disadvantages of MRI

Extreme precautions must be taken to keep metallic objects out of the room where the machine is operating People with pacemakers can't safely be scanned Some people suffer from claustrophobia, and find the confinement discomforting The machine makes a very loud continuous hammering noise when operating Some people are too big to fit inside the magnet MRI scans require patients to hold very still for long periods of time ... up to 90 minutes or more in some cases MRI systems are expensive to buy and run so are currently beyond most DGHs. An MRI scan costs about 500 to the NHS

MRI installed base

1990 unit sales of MRI systems, tens to hundreds of MRI scans 2004 installed base is 12,000 MRI systems 75-80 million scans per year (400/scan) Market growth at 10% pa Major growth is in high field MRI ( 3T) Despite much excitement about open magnet systems, take-off is slow (Hitachi dominate this sector) Special purpose MRI systems have not had much impact (sigh)

Energy (spin) states

Long before Tony Blair, Quantum Mechanics invented the concept of spin In a strong magnetic field, nuclei act like tiny dipole magnets that align with or, amazingly, against the field

Strong magnetic field

B0

Aligned with the field: lower energy state

Aligned against the field: higher energy state

Protons (= hydrogen nuclei) have two spin states. They act like miniature tops and precess about the field direction

Spin, moment
All nuclei have spin multiples of Combined with charge moment Nucleus with odd spin acts like a small dipole magnet If nucleus has S spin states, the moment (magnet) has 2S+1 stable states in an external magnetic field Hydrogen (proton): S = 2 states

Alignment of Spins in a Magnetic Field

M
magnetic moment

M=0

spin

With no magnetic field, the spins are randomly aligned

B0 field

Energy in a Magnetic Field (Zeeman Splitting, Spin )

mI = +

E+1/2= B0/2 P+1/2= 0. 5000049

mI =

E-1/2= +B0/2 P-1/2= 0.4999951

N up
1.5T, T=310K, P(E)exp(E/kT)

N down

hB0 exp kT

Excess of protons aligned with field

For an external magnetic field of 3T, there are only about 10 per million more protons parallel to the field than anti-parallel!! Nevertheless, there are millions of protons, so this is enough to give a useful magnetic field The smaller the field, the fewer the excess, the poorer the SNR so use a very large magnetic field

NMR the key discovery

The nucleus is bombarded with Radiofrequency (RF) energy An unsuspecting low energy nucleus At certain resonant frequencies of the RF energy, the proton flips to the high energy state

When the RF energy is turned off, the newly high energy nucleus may revert to its low energy state, giving off RF energy in the process

So what?
The resonant frequency at which this happens is called the Larmor frequency: In this equation,

= B0

is the gyromagnetic constant for the stuff that is being energised

Critically, this constant depends on the biochemical nature of the stuff and its surroundings: Chemical Shift A typical field strength B0 used in MRI is 1.5 Tesla At this field strength, the Larmor frequencies for Hydrogen and Carbon 13 (the atoms most relevant in medical imaging) are 63.9 MHz and 16.1 MHz respectively. Probing with different frequencies of RF energy enables us to build a spectrum of what is in the sample

Exciting a spin system

subject it to a short period of high intensity radiowaves at a frequency close to the Larmor frequency This is called the B1 field, orientated in a direction perpendicular to, and rotating about, the B0 field. The magnitude of B1 10-5 B0 In a co-ordinate system rotating at or close to the Larmor frequency, this results in rotation of the magnetization away from the direction of the external magnetic field

B1 rotates about B0

B0

Applying a pulse B1
z-axis

B0

Mz

Resultant magnetic field on the voxel

The longer the RF pulse is applied, and the stronger it is, the bigger the deflection of the net magnetic field, that is, the bigger the angle . It can reach 90, or even 180 degrees. The bigger , the longer it takes to recover when the RF is turned off.

M xy
B1

x-y plane

Free Induction Decay

FT

time frequency

FT
time frequency

The relaxation constant T1 For the nuclei to return to their initial energy states by emitting energy (the MR signal), the excited spin system must be exposed to an electromagnetic field oscillating with a frequency at or close to the Larmor frequency. This process can occur by the nuclei being `stimulated' by surrounding nuclei and is assumed to occur in a simple exponential manner

M z (t ) M 0 = (M z (0) M 0 )e

t T1

T1 is called the spin-lattice relaxation time It corresponds to the time required for the system to return to 63% of its equilibrium value after it has been exposed to a 90 pulse

T1 Weighted Imaging
white matter

grey matter
Contrast

T1b ln T1a T1b T1a Optimal TR = T1a T1b

Optimal TR

TR

T1 Weighted Image

T1/s white matter grey matter CSF 0.7 1 4 1.5T

R1/s-1 1.43 1 0.25

SPGR, TR=14ms, TE=5ms, flip=20

T1 Relaxation
dMz(t) = [Mz(t) M0] dt T1 Mz(t) = M0 + {Mz(0) M0}exp(-t/T1)
saturationrecovery Mz Mz(0) = 0 t M0 Mz inversionrecovery M0

Mz(0) = M0 t

The contribution of all the spins precessing around the external magnetic field B0 produces a net magnetisation M0. When a 90 RF pulse is applied, this net magnetisation is tipped onto the x,y-plane. Dephasing of the spins results in a quick decrease of the net magnetisation in the x,y-plane. The dephasing is exponential and characterised by T2.

The relaxation constants T2 T2

Immediately after a pulse is applied, all of the nuclei precess around the magnetic field in phase. As time passes, the spins begin to dephase and so the observed signal decreases. They do so according to:

M T (t ) = M T (0)e

T2

T2 is called the spin-spin relaxation time T2 values are 40-200ms depending on the tissue T2 is approximately ten times smaller than T1. Different scan sequences show up differences in these relaxation times generating what are referred to as T1, T2 or proton density (the concentration of protons) weighted images.

T2 Relaxation
dMxy(t) = Mxy(t) dt T2 Mxy(t) = Mxy(0) exp(t/T2)
Mxy

T2 Weighted Imaging
Echo Amplitude

grey white
Contrast

T2 a T2 a T2 b ln T2 b Optimum TE = T2 a T2 b

Optimum TE

TE

T2 Weighted Image

T2/ms CSF grey matter white matter 1.5T 500 8090 7080

SE, TR=4000ms, TE=100ms

Slice Selection
time G 0 frequency

To image a slice of material requires a method of exciting only material within that slice. This is achieved by superimposing a small spatially varying magnetic field, Gz , called a gradient field. The field is applied in the same direction as while the RF pulse is applied.

Recall Larmor' s equation : = B0

+ ( x ) = (B0 + x.B ) , so knowing B and the chemical shift enables us to estimate x

Now suppose there is a magnetic field gradient B per metre

Three mutually orthogonal gradient coils are used to localise a nucleus in x-, y-, and z-. Normally a slice is localised (z) and then a coil pair in x,y

MRI: the Fourier Transform

FFT

K-space as measured in the MRI experiment

The medical image we inspect is the FT of k-space

Note: motion of the subject will be local in k-space so have a global effect on the image!

K-space

Each row of k-space contains the raw data received under a particular phase gradient, where the order in which the rows are recorded depends on the imaging sequence used; Once all of kspace has been assembled, it is Fourier transformed (2D FFT) to obtain the image

Full k Space Coverage

ky

kx

Only Centre of k Space

ky

kx

Only Edges of k Space

ky

kx

Illustration of the MR read gradient and signals generated at different spatial locations. The illustration shows all the signals in phase which corresponds to the zeroth row of k-space. The material at A and D has a high MR signal, the material at B and C has a low signal. A and C are towards the left so have a low frequency of precession; B and D are on the right so have a higher frequency.

phase-encoding gradients

A, B and C have high, medium and low MR signals. They are at the same x-position so have the same frequency of precession. Their y-positions are encoded by repeated scanning with different phase gradients. With a zero phase gradient (the central row), all the signals are in phase. With a positive phase gradient (the top row), A has a phase lead, and C has a phase lag with respect to B. Recording the signals under all phase gradients allows the y-positions to be recovered by Fourier Transforming k-space.

Why Use k Space?

Larmor equation = B0
x=1cm x=0 x=+1cm

(x,y) = B0 + Gxx + Gyy phase (x,y,t) = 2 B0dt + 2 Gxxdt + 2 Gyydt elemental signal S(x,y,t) = (x,y) exp{i (x,y,t)} total signal S(t) = (x,y) exp{i (x,y,t)} dxdy

A Few Substitutions
From:

total signal S(t) = (x,y) exp{i (x,y,t)} dxdy kx(t) = Gxdt ky(t) = Gydt

In rotating frame: To:

total signal S(t) = (x,y) exp{2i(kxx+kyy)} dxdy This is the standard Fourier Equation!

A potted history of MRI 1: NMR

Felix Bloch & Edward Purcell (1946) NP52 Mostly uses oxygen and potassium nuclei eg ATP consumption in the heart Oxford has made massive contributions Rex Richards, Walter Bodmer, George Radda, Continues for spectroscopy currently 18T systems for protein structure (OI/Brucker) NMR says that there is a substance present in a Many of the terms compound but cant say where
Freq. occurrence

and concepts in MRI originate in NMR:

T1 , T2 , flip angle, pulse sequence, ...

Molecular weight

A potted history of MRI

1973 Paul Lauterbur demonstrated MRI using the back projection ideas introduced in CT (the same year) NP 03 1975 Richard Ernst introduced the modern technique kspace NP 91 1977 Peter Mansfield introduced EPI basis for all modern fast imaging & fMRI NP 03 1980 first whole body image, 1986 first clinical systems, 1987 first research cardiac motion sequence, 1989 Siemens OI joint venture to make superconducting magnets, 1989-1997 competitive advantage based on technological superiority; 1998 MR system becomes commoditized competitive advantage based on systems integration & software