Presentation is loading. Please wait.

Presentation is loading. Please wait.

Unbiased Longitudinal Processing of Structural MRI in FreeSurfer Martin Reuter

Similar presentations


Presentation on theme: "Unbiased Longitudinal Processing of Structural MRI in FreeSurfer Martin Reuter"— Presentation transcript:

1 Unbiased Longitudinal Processing of Structural MRI in FreeSurfer Martin Reuter

2 What can we do with FreeSurfer? measure volume of cortical or subcortical structures compute thickness (locally) of the cortical sheet study differences of populations (diseased, control)

3 Why longitudinal? to reduce variability on intra-individual morph. estimates to detect small changes, or use less subjects (power) for marker of disease progression (atrophy) to better estimate time to onset of symptoms to study effects of drug treatment... We'd like to: exploit longitudinal information (same subject, different time points))

4 Challenges in Longitudinal Designs Over-Regularization: Temporal smoothing Non-linear warps Limited designs: Only 2 time points Special purposes (e.g. only surfaces, only CSF/WM/GM) Bias [Reuter and Fischl 2010] Interpolation Asymmetries [Yushkevich et al 2010] Asymmetric Registrations Asymmetric Information Transfer

5 How can it be done? Stay unbiased with respect to any specific time point by treating all the same Create a within subject template (base) as an initial guess for segmentation and reconstruction Initialize each time point with the template to reduce variability in the optimization process For this we need a robust registration (rigid) and template estimation

6 Robust Registration [Reuter et al 2010] Inverse consistency: a symmetric displacement model: resample both source and target to an unbiased half-way space in intermediate steps (matrix square root) SourceTargetHalf-Way

7 Robust Registration [Reuter et al 2010] (c.f. Nestares and Heeger, 2000)‏ Square Tukey's Biweight Limited contribution for outliers [Nestares&Heeger 2000]

8 Robust Registration [Reuter et al 2010] Tumor data courtesy of Dr. Greg Sorensen Tumor data with significant intensity differences in the brain, registered to first time point (left).

9 Robust Registration [Reuter et al 2010] Target

10 Robust Registration [Reuter et al 2010] Registered Src FSL FLIRTRegistered Src Robust

11 mri_robust_register mri_robust_register is part of FreeSurfer can be used for pair-wise registration (optimally within subject, within modality) can output results in half-way space can output ‘outlier-weights’ see also Reuter et al. “Highly Accurate Inverse Consistent Registration: A Robust Approach”, NeuroImage for more than 2 images: mri_robust_template

12 Robust Template Estimation Minimization problem for N images: Image Dissimilarity: Metric of Transformations:

13 Robust Template Intrinsic median: -(Initialization) -Create median image -Register with median (unbiased) -Iterate until convergence Command: mri_robust_template

14 Robust Template: MEDIAN Top: difference Left: mean Right: median The mean is more blurry in regions with change: Ventricles or Corpus Callosum Median: -Crisp -Faster convergence -No ghosts

15 Robust Template for Initialization Unbiased Reduces Variability Common space for: - TIV estimation - Skullstrip - Affine Talairach Reg. Basis for: - Intensity Normalization - Non-linear Reg. - Surfaces / Parcellation

16 Bias Biased information transfer: [BASE1] and [BASE2]. Our Method: [FS5.1] [FS5.1rev] shows no bias. SubcorticalCortical

17 Simulated Atrophy (2% left Hippo.) Cross sectional RED, longitudinal GREEN Simulated atrophy was applied to the left hippocampus only Left HippocampusRight Hippocampus

18 Test-Retest Reliability [LONG] significantly improves reliability 115 subjects, ME MPRAGE, 2 scans, same session SubcorticalCortical

19 Test-Retest Reliability [LONG] significantly improves reliability 115 subjects, ME MPRAGE, 2 scans, same session Diff. ([CROSS]-[LONG]) of Abs. Thick. Change: Significance Map

20 Increased Power Sample Size Reduction when using [LONG] Left Hemisphere:Right Hemisphere

21 Huntington’s Disease (3 visits) [LONG] shows higher precision and better discrimination power between groups (specificity and sensitivity). Independent ProcessingLongitudinal Processing

22 Huntington’s Disease (3 visits) Putamen Atrophy Rate can is significant between CN and PHD far, but baseline volume is not. Rate of AtrophyBaseline Vol. (normalized)

23 Still to come … Future Improvements: Same voxel space for all time points (in FS 5.1) Common warps (non-linear) Intracranial volume estimation Joint intensity normalization New thickness computation Joint spherical registration Thanks to: the FreeSurfer Team, specifically Bruce Fischl and Nick Schmansky


Download ppt "Unbiased Longitudinal Processing of Structural MRI in FreeSurfer Martin Reuter"

Similar presentations


Ads by Google