Detecting Subtle Changes in Structure l Chris Rorden

Detecting Subtle Changes in Structure l Chris – Rorden Diffusion Tensor Imaging l Measuring white matter integrity l Tractography and analysis. 1

Diffusion Weighted Imaging l Diffusion weighted images measures random motion of water molecules. – – In ventricles, CSF is unconstrained, so high velocity diffusion In brain tissue, water more constrained, so less diffusion. DWI 2

Diffusion Tensor Imaging (DTI) l l DTI is an extension of DWI that allows us to measure direction of motion. DTI allows us to measure both the velocity and preferred direction of diffusion – – In gray matter, diffusion is isotropic (similar in all directions) In white matter, diffusion is anisotropic (prefers motion along fibers). 3

DTI l l The amount of diffusion occurring in one pixel of a MR image is termed the Apparent Diffusion Coefficient (ADC) or Mean Diffusivity (MD). The non-uniformity of diffusion with direction is usually described by the term Fractional Anisotropy (FA). MD differs FA differs 4

Raw DTI data DTI scans apply gradients of different strengths (b-values) and directions (b-vectors). l These highlight different diffusion directions. l Simplest DTI is shown below: one b=0 image (no directionality), plus six images with b=1000 but each with a different b-vector. l 5

Eddy current correction l l l Gradients for DTI data cause spatial distortions. Different directions have different distortions. The B 0 image has little distortion (it is a T 2 weighted Image) Eddy current correction aligns all directions to the b 0 (reference) image. Analogous to motion correction for f. MRI data. 6

What is a tensor? l A tensor is composed of three vectors. – l l Think of a vector like an arrow in 3 D space – it points in a direction and has a length. The first vector is the longest – it points along the principle axis. The second and third vectors are orthogonal to the first. Sphere: V 1=V 2=V 3 Football: V 1>V 2 V 1>V 3 = V 2 ? ? ? : V 1>V 2>V 3 7

DTI MD FA V 1 modulated by FA Color shows principle tensor direction, brightness shows FA 8

The crossing fibers problem l Tensor (DTI) will have trouble distinguishing voxels with crossing fibers from isotropic regions. Reality Tensors CSF (isotropic) Crossing fibers 9

Tractography l Programs like med. Inria allow us to measure integrity of connections between different regions. 10

Statistics – MD and FA The FA and MD maps from each individual can be normalized to standard space. l Standard voxelwise statistics applied (like VBM) l Allows us to infer differences in white matter integrity between groups. l Group mean images based on normalized data. 11

Statistics – tensors You can measure fiber strength connectivity brain regions and see if this differs between groups. l Traditionally, tensor maps are very difficult to normalize – we do not want to squish relative shape of tensor. l – Recent advances are addressing this (DTITK) 12

TBSS - Tract-Based Spatial Statistics l l l TBSS is a FSL approach to conduct between group comparisons of DTI data. projects data onto group-mean tract skeleton, allowing voxelwise analysis addresses alignment problems unsolved by nonlinear registration Overview www. fmrib. ox. ac. uk/fsl/tbss/index. html Tutorial www. fmrib. ox. ac. uk/fslcourse/lectures/practicals/fdt/index. htm 13

Sample application Young adults vary in their ability to recollect information. l TBSS shows that Fornix integrity predicts this variability (but not variability in familiarity). l Rudebeck (2009) Jo. N 14