introduction to freesurfer

1

Introduction to FreeSurferhttp://surfer.nmr.mgh.harvard.edu

2

Post Your Questions!

http://surfer.nmr.mgh.harvard.edu/cgi-bin/fsurfer/questions.cgi

3

Why FreeSurfer?

• Anatomical analysis is not like functional analysis – it is completely stereotyped.

• Registration to a template (e.g. MNI/Talairach) does not account for individual anatomy.

• Even if you do not care about the anatomy, anatomical models allow functional analysis not otherwise possible.

4

ICBM Atlas

12 DOF(Affine)

Why not just register to an ROI Atlas?

5

Subject 1Subject 2 aligned with Subject 1

(Subject 1’s Surface)

Problems with Affine (12 DOF) Registration (you will get sick of this slide)

6

Surface and Volume Analysis

Cortical Reconstructionand Automatic Labeling

Inflation and FunctionalMapping

Surface FlatteningSurface-based Inter-subjectAlignment and Statistics

Automatic SubcorticalGray Matter Labeling

Automatic Gyral White Matter Labeling

7

Talk Outline

1. Cortical (surface-based) Analysis.

2. Volume Analysis.

8

Talk Outline

1. Cortical (surface-based) Analysis.

2. Volume Analysis.

9

Why Is a Model of the Cortical Surface Useful?

Local functional organization of cortex is largely 2-dimensional! Eg, functional mapping of primary visual areas:

From (Sereno et al, 1995, Science). Also, smooth along surface

10

Flat Map of Monkey Visual Areas

D.J. Felleman and D.C. Van Essen, CC, 1991

11

What Can One Do With A Surface Model?

left primary visual cortex right visual hemifield

desired shape of activity pattern required shape of stimulus

goal: use model to impose desired activity pattern on V1

Collaboration with Jon Polimeni and Larry Wald.

w=k log(z+a)

12

Tangential Resolution Measured with Surface-based Analysis

Collaboration with Jon Polimeni and Larry Wald.

13

Tangential Resolution Measured with Surface-based Analysis

Collaboration with Jon Polimeni and Larry Wald.

14Thanks to Larry Wald for this slide.

M-G-H

15

Surfaces: White and Pial

16

Inflation

17

Inflated surface with cuts superior temporal

Metrically optimal flat map

calcarine

central

sylvian

anterior

posterior

Surface Flattening – Whole Hemisphere

18

Surface Model

• Mesh of triangles gives a measurable size

• Allows us to measure Area, Curv., Thickness (distance b/w vertices)

• Vertex = point of 6 triangles

• Triangles/Faces ~ 150,000 per hemi

• 1:1 correspondence of vertices

• XYZ at each vertex

19

Cortical Thickness

white/gray surface

pial surface

lh.thickness, rh.thickness

• Distance between white and pial surfaces

• One value per vertex

20

A Surface-Based Coordinate System

21

Comparing Coordinate Systems and Brodmann Areas

Cumulative histogram (red=surface,

blue=nonlinear Talairach)

Ratio of surface accuracy to volume accuracy

22

Automatic Surface SegmentationPrecentral

Gyrus Postcentral Gyrus

Superior Temporal GyrusBased on individual’s folding pattern

23

Inter-Subject AveragingS

ubje

ct 1

Sub

ject

2

Native Spherical Spherical

Surface-to- Surface

Surface-to- Surface

GLM

Demographics

mri_glmfit cf. Talairach

24

Borrowed from (Halgren et al., 1999)

Visualization

25

Rosas et al., 2002

Kuperberg et al., 2003

Gold et al., 2005

Rauch et al., 2004Salat et al., 2004

Fischl et al., 2000

Sailer et al., 2003

26

Talk Outline

1. Cortical (surface-based) Analysis.

2. Volume Analysis.

27

Volume Analysis: Automatic Individualized Segmentation

• Surface-based coordinate system/registration appropriate for cortex but not for thalamus, ventricular system, basal ganglia, etc…

• Anatomy is extremely variable – measuring the variance and accounting for it is critical (more in the individual subject talk)!

28

Volumetric Segmentation (aseg)

Caudate

Pallidum

Putamen

Amygdala

Hippocampus

Lateral Ventricle

Thalamus

White Matter

Cortex

Not Shown:Nucleus AccumbensCerebellum

29

Volume Differences Predictive of AD

LH RH0

0.5

1

1.5

2

2.5

blue=ctrl (25), cyan=questbl (71), y=converters (21), red=AD (17)

late

ral-v

entr

icle

vol

ume

(pct

bra

in)

Data courtesy of Drs Marilyn Albert and Ron Killiany

30

Combined Segmentation

aparc+aseg

aseg

aparc

31

Gyral White Matter Segmentation

Nearest Cortical Labelto point in White Matter

wmparc

+ +

aparc

aparc+aseg

32

Summary• Why Surface-based Analysis?

– Function has surface-based organization– Visualization: Inflation/Flattening– Cortical Morphometric Measures– Inter-subject registration

• Automatically generated ROI tuned to each subject individually

Use FreeSurfer Be Happy

33

MGHBruce Fischl

Allison StevensNick Schmansky

Andre van der KouweDoug GreveDavid Salat

Evelina BusaLilla Zöllei

Koen Van LeemputSita KakunooriRuopeng Wang

Rudolph Pienaar Krish Subramaniam

Diana Rosas

Acknowledgements

UC San DiegoAnders Dale

MITPolina Golland

B. T. Thomas YeoMert Sabuncu

Florent SegonnePeng Yu

Ramesh Sridharan

MGHJean Augustinack

Martin ReuterAnastasia Yendiki

Jon PolimeniKristen Huber

UCLMarty Sereno

NINDS

MGH (past)Brian T Quinn

Xiao HanNiranjini Rajendran

Jenni Pacheco

Sylvester CzannerGheorghe Postelnicu

Sean Marrett