The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital Diffusion weighted MRI Brian Hansen, PhD brianh@phys.au.dk Lecture outline • Background: The physics of diffusion Fickian diffusion Brownian motion Self diffusion • Diffusion measurements PGSE pulse sequence and spin dynamics Interpreting the diffusion weighted signal Diffusion in biological tissues • Diffusion MRI in neuroimaging / neuroscience Diffusion weighted MRI and the ADC map Other applications: Fibre tracking The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 2 Background and motivation • Diffusion processes are everywhere in Nature Gases Solids (semi-conductors, alloys) Liquids (chemical reactions, biology, physiology) • Diffusion Weighted (DW) MRI is a non-invasive method for measuring diffusion: Diffusion coefficient (physical or apparent) Direction of diffusion (preferred direction) • From these parameters the state of e.g. tissue can be estimated. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 3 Clinical Application • • Ischemic infarction is not visible on conventional MRI (T1,T2, PD) DW MRI introduces new sensitivity: The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 4 Physical Principles The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 5 Fickian diffusion Fick’s two laws describe diffusion driven by a difference in concentration. C(x,t) Fick’s 1st: Fick’s 2nd: x The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 6 Self-diffusion • All water molecules perform a thermally driven random walk. • We can only describe this motion statistically: For Brownian motion z = 2 The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital H2O Brownian Motion • • • Named after scottish botanist Robert Brown (1773-1858). Explained by Einstein in 1905. The thermal motion of the molecules cause them to collide. Random motion follows. Described by the Stokes-Einstein relation: D is diffusion coefficient, kB is the Boltzmann constant, T is absolute temperature, m is liquid viscosity and r is particle radius. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 8 Diffusion in biological tissue Diffusion in tissue is resticted by cell membranes, organelles etc: These random trajectories will in time fill the plane and reveal the structure. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital Measuring Diffusion • The Pulsed Gradient Spin Echo (PGSE) sequence 180 90 g D time The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 10 No diffusion: Stationary spins are unaffected by diffusion gradient. time Spin 1 Spin 2 Spin 3 B x The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 11 With diffusion: time Spin 1 Spin 2 Spin 3 B x The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 12 Vector sum + + = No loss of signal in areas with no diffusion. Diffusion introduces a signal loss. + + = High diffusion gives strong signal attenuation. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 13 DW MRI parameters 180 90 g Parameters g, , D are combined in the b-factor: D Here g is the proton gyromagnetic ratio. The b-factor can be varied by varying one of g, and D. For the PGSE sequence the case b = 0 corresponds to the simpel SE sequence. A large b-factor gives a large signal loss in areas with high diffusion. This is called strong diffusion weighting. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 14 Signal and b-factor I The DWMR signal from simple free diffusion is described by: By ”simple free diffusion” we mean that applies for all times. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 15 Signal and b-factor II On a log-plot this yields a straight line: log(S(b)/S(b=0)) = -bD The slope of the curve gives us the physical diffusion coefficient. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 16 Measuring D Remember: The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 17 Diffusion in tissue Diffusion in biological tissue is not free: Simple signal behaviour breaks down due to complex tissue structure: Cell membranes, organelles etc. restrict the diffusion of the water molecules. This means that no longer applies for all times. Grey matter, ECS in red. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 18 The Apparent Diffusion Coefficient • • • The value we measure is no longer the physical diffusion coefficient: Instead we get an average over many restricted random walks We introduce the term Apparent Diffusion Coefficient (ADC) Two measurements at b = 0 and b = 1000 s/mm2 are made: The slope gives the ADC – not the physical diffusion coefficient. Typical ADC values in brain (mm2/s): Normal gray matter: 0.8-1.010-3 Normal white matter: 0.2-1.0 10-3 Free water (CSF): 2.9 10-3 Review and references in Journal of Computer Assisted Tomography 25(4):515-519. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 19 Increasing b-values b = 2000 b = 4000 b = 7000 The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 20 DWI and ADC maps DW MRI provides two new image types: The Diffusion Weighted Image (DWI) The ADC map (a calculated image) The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 21 Stroke DWI Acute DWI tPA + 2h DWI tPA + 24h DWI The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 22 Possible cause of the bright areas in the DWI: Cells in normal tissue The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital Cells in infarct (stroke) Summary: Strength of DWI T2 MRI: DWI: • Infarct is not visible – brain appears normal • Infarct clearly visible • Scan time: 30 sec, EPI • No IV contrast agent needed • Infarct detectable after few minutes The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 24 Summary: Image types Two image types: DWI and ADC map: DWI: ADC map: • Signal is diffusion weighted. • Calculated image • High diffusion: signal loss • Contrast opposite to DWI • Low diffusion: no signal loss • Low intensity: low ADC value (low • Infarcts are bright diffusion) • High intensity: high ADC (high diffusion) The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 25 DWI ADC MTT Osvd The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 26 Diffusion Tensor Imaging Diffusion is often directional – e.g. along fibers: Instead of measuring many b-values we measure along many different directions. Instead of the ADC we obtain the Diffusion Tensor which describes the diffusion coefficient in space. This is the basis of fibre tracking. The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital 27 Images courtesy of Jesper Frandsen, CFIN The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital Image courtesy of Jesper Frandsen, CFIN The Danish National Research Foundation’s Center of Functionally Integrative Neuroscience Aarhus University / Aarhus University Hospital