Richard Wise FMRI Director wiserg@cardiff.ac.uk +44(0)20 2087 0358 FMRI acquisition Richard Wise FMRI Director wiserg@cardiff.ac.uk +44(0)20 2087 0358

I am going to introduce you to the methods with the help of … John and Rami … say hello

Why do we need the magnet?

Inside an MRI Scanner z gradient coil r.f. transmit/receive x gradient coil Why do we need a magnet and what are all these complicated coils inside? super conducting magnet subject gradient coils

Common NMR Active Nuclei Isotope Spin % g I abundance MHz/T 1H 1/2 99.985 42.575 2H 1 0.015 6.53 13C 1/2 1.108 10.71 14N 1 99.63 3.078 15N 1/2 0.37 4.32 17O 5/2 0.037 5.77 19F 1/2 100 40.08 23Na 3/2 100 11.27 31P 1/2 100 17.25

Nuclear Spin M magnetic moment M=0 spin If a nucleus has an unpaired proton it will have spin and it will have a net magnetic moment or field

Resonance If a system that has an intrinsic frequency (such as a bell or a swing) can draw energy from another system which is oscillating at the same frequency, the 2 systems are said to resonate

Spin Transitions High energy Low energy Show RF coil

ω = γ B The Larmor Frequency Frequency  Field strength Demo FID Modify B0 to show change in frequency 128 MHz at 3 Tesla

Tissue magnetization B0 M 90º RF excitation pulse Field strength - the stronger the field the bigger is M More signal - more blobs

Tissue magnetization B0 M 90º RF excitation pulse MR signal ω = γ B

Tissue magnetization B0 M 90º RF excitation pulse . MR signal ω = γ B

Tissue magnetization B0 90º RF excitation pulse MR signal ω = γ B Demo FID signal Signal decay: time constant T2 time

Tissue contrast: TE &T2 decay Echo Amplitude Long T2 (CSF) Medium T2 (grey matter) Explain that we really acquire an echo rather than a single FID Echo is needed to get more complete artifact free data Also allows us to add interesting contrast such as BOLD and diffusion. Contrast Short T2 (white matter) TE

T2 Weighted Image

T2 Weighted Image T2/ms CSF 500 grey matter 8090 white matter 7080 SE, TR=4000ms, TE=100ms SE, TR=4000ms, TE=100ms

Tissue magnetization B0 M M Magnetization recovery: time constant T1 Explain TR - repeat time determines T1 contrast M Magnetization recovery: time constant T1 time

Tissue magnetization B0 M M Magnetization recovery: time constant T1

Tissue contrast: TR & T1 recovery Short T1 (white matter) Mz Medium T1 (grey matter) Long T1 (CSF) Contrast TR

T1 Weighted Image SPGR, TR=14ms, TE=5ms, flip=20º Normally acquired for your FMRI structural scan SPGR, TR=14ms, TE=5ms, flip=20º

T1 Weighted Image T1/s R1/s-1 white matter 0.7 1.43 grey matter 1 1 CSF 4 0.25 1.5T SPGR, TR=14ms, TE=5ms, flip=20º SPGR, TR=14ms, TE=5ms, flip=20º

Short TR Long TR Short TE Long TE T1 PD T2 Signal comes principally from water Depends on tissue properties Amount of water Physical and chemical structure Contrast comes from ‘relaxation’ processes

From Frequencies to Images Vary the field by position Decode the frequencies to give spatial information

Gradient coils z gradient coil r.f. transmit/receive x gradient coil Why do we need a magnet and what are all these complicated coils inside? super conducting magnet subject gradient coils

Image formation Fourier Transform time frequency Signal Spectrum

The Fourier Transform FFT n 2 x 2 Matrix size (2 to the power n) for the convenience of the fast fourier transform 2 x 2 n

Slice selection RF excitation ω = γ B time frequency 0 G

(Gradient echo) Pulse sequence Pulse sequence allows us to spatially encode in 2 or 3 dimensions

The Pulse Sequence Controls Slice location Slice orientation Slice thickness Number of slices Image resolution Field of view (FOV) Image matrix Echo-planar imaging Image contrast TE, TR, flip angle, diffusion etc Image artifact correction Saturation, flow compensation, fat suppresion etc

T2* : pleasure …..

T2* : ….. and pain

T2* contrast

T2* contrast Field variation across the sample Decay of summed NMR signal

GE-EPI is T2* weighted

Wilson et al Neuroimage 2003 CONTRAST T1 spin lattice relaxation time Thermal processes How quickly the system gets back to equilibrium T2 spin spin relaxation Loss of coherence in transverse magnetisation Random fluctuation in magnetic field felt by nucleus T2* macroscopic field gradients Loss of coherence across sample Optimum TE = T2* Wilson et al Neuroimage 2003

Neural activity to FMRI signal Signalling Vascular response Vascular tone (reactivity) Autoregulation Metabolic signalling BOLD signal glia arteriole venule B0 field Synaptic signalling Blood flow, oxygenation and volume

FMRI and electrophysiology Logothetis et al, Nature 2001

Haemodynamic response balloon model Functional contrast (and what each is good for) BOLD (blood oxygenation level dependent) T2 and T2* Perfusion (arterial spin labelling) - more quantitative but lower SNR Cerebral blood volume VASO, vascular space occupancy Functional Spectroscopy % -1 initial dip undershoot Buxton R et al. Neuroimage 2004

Field inhomogeneity disrupts magnetic field and reduces image quality.

Blood oxygenation Bandettini and Wong. Int. J. Imaging Systems and Technology. 6:133 (1995) Bandettini and Wong. Int. J. Imaging Systems and Technology. 6:133 (1995)

Active: 40% increase in CBF, 20% increase in CMRO2 Rest O2 Sat 100% 80% 60% O2 O2 O2 Active: 40% increase in CBF, 20% increase in CMRO2 BOLD signal comes mainly from the venous side O2 Sat 100% 86% 72% CMRO2 = OEF  CBF

CMRO2: CBF ratio Hoge R et al Ant insula was most significantly modulated – consistent with previous imaging reports. Hoge R et al

Signal evolution R2*  (1-Y) CBV S = Smax . e-TE/R2* Deoxy-Hb contribution to relaxation R2*  (1-Y) CBV Y=O2 saturation b~1.5 Gradient echo S = Smax . e-TE/R2* Longer TE, more BOLD contrast but less signal and more dropout/distortion. TE=T2*

Harrison RV et al. Cerebral cortex. 2002 Vessel density 500 m 100 m Harrison RV et al. Cerebral cortex. 2002

Resolution Issues Spatial Resolution Temporal Resolution How close is the blood flow response to the activation site (CBF better?) Most BOLD signal is on the venous side EPI is “low res” Dropout and distortion Slice orientation Slice thickness Temporal Resolution

Factors affecting BOLD signal? Physiology Cerebral blood flow (baseline and change) Metabolic oxygen consumption Cerebral blood volume Equipment Static field strength Field homogeneity (e.g. shim dependent T2*) Pulse sequence Gradient vs spin echo Echo time, repeat time Resolution BOLD signal is inherently messy being a mixture of physiological changes. It is not directly comparable between brain regions as it depends on vessel architecture but we see later on that this may be a benefit in yielding information about cerebrovasculature.

Physiological baseline Baseline CBF, But CBF CMRO2 unchanged (Brown et al JCBFM 2003) BOLD response  Implications for disease and phsyiological state, I.e. breathing and sedation, anxiety Vasoactive medications Cohen et al JCBFM 2002

Noise sources What is noise in a BOLD experiment? Unmodelled variation in the time-series Intrinsic MRI noise Independent of field strength, TE Thermal noise from subject and RF coil Physiological noise Increases with field strength, depends on TE At 3T physiological noise > intrinsic Cardiac pulsations Respiratory motion and B0 shift Vasomotion, 0.1Hz Blood gas fluctuations “Resting state” networks Also Scanner drift (heating up) Noise is why we are here

BOLD Noise structure 1/f dependence BOLD noise BOLD is bad for detecting long time-scale activation frequency

Spatial distribution of noise Motion at intensity boundaries Head motion Respiratory B0 shift Physiological noise in blood vessels and grey matter

Thanks to … John Evans Rami Niazy Martin Stuart Spiro Stathakis