Introduction / Overview 8th October 2008 Hanneke den Ouden, Justin Chumbley, Maria Joao Rosa Wellcome Trust Centre for Neuroimaging, UCL

Introduction / Overview8th October 2008

Hanneke den Ouden, Justin Chumbley, Maria Joao Rosa

Wellcome Trust Centre for Neuroimaging, UCL

Overview

• Introduction

• What’s MfD

• Programme for 2008

• How to prepare your presentation

• Where to find information and help

• Experts

• Overview for dummies

Introduction to MfD 2008

Methods for Dummies 2008

• Basic Statistics

• fMRI (BOLD)

• EEG / MEG

• Connectivity

• VBM & DTI


Areas covered in MfD

Wednesdays / 13h00 – 14h00 / FIL Seminar Room

Aim: to give a basic introduction to human brain imaging analysis methods, focusing on fMRI and M/EEG

PROGRAMME 2008

Autumn


I. Basic Statistics 15th Oct – 12st Nov

• Linear Algebra & Matrices (Nicholas Wright, Nick Henriquez)

• T-tests, ANOVA’s & Regression (Nicholas Wright, Lorelei Howard)

• General Linear Model (Ramiro Joly, Sinead Mullally)

• Bayes for beginners (Stephen Fleming, Sharon Gilaie-Dotan)

• Random Field Theory (Christian Lambert, Cirian Hill)


II. What are we measuring? 19th Nov – 26st Nov

• Basis of the BOLD signal (Christoph Korn, Andrea Dantas)

• Basis of the M/EEG signal (Bonnie Breining, Evelyne Mercure)


III. fMRI Analysis3th Dec – 17rd Dec

• Preprocessing:

– Realigning and un-warping (Mark Weyers, Hanna Marno)

– Co-registration & spatial normalisation (Catherine Sebastian, Antoinette Nicolle)

• Study design and efficiency (Nicholas Wright, Edoardo Zamuner)


Continues after Christmas break…

PROGRAMME 2008

Spring



• 1st & 2nd level analysis – Design matrix contrasts and inference (Tessa Decker &

Emmanuelle Volle)

• Parametric modulation, temporal basis functions and correlated regressors (Mkael

Symmonds, Patrick Freund)

III. fMRI Analysis (cont.)14th Jan – 21st Jan

IV. EEG & MEG28th Jan – 4th Feb

• Pre-processing and experimental design (Nicolas Abreu, Mathias Gruber)

• Contrasts, inference and source localisation (Maro Machizawa, Himn Sabir)


V. Connectivity 11th Feb – 25th Feb

• Intro to connectivity - PPI & SEM (Karine Gazarian, Carmen Tur)

• DCM for fMRI – theory & practice (Nikos Konstantinou, Stephanie Burnett)

• DCM for ERP / ERF – theory & practice (Giovanna Moretto, Saloni Krishnan)



VI. Structural MRI Analysis 4nd Mar & 11th Mar

• Voxel Based Morphometry (Thomas Doke, ChiHua Chen)

• Basics of DTI (Nikos Gorgoraptis, Rohit Khanna)

How to prepare your presentation

• Remember your audience are not experts…

• The aim of the sessions is to

– introduce the concepts and explain why they are important to imaging analysis

– familiarise people with the basic theory and standard methods

• Time: 45min. + 15min. questions – 2 presenters per session

• Don’t copy last year’s slides!!!...

• Start preparing your talk with your co-presenter at least 2 weeks in advance

• Talk to the allocated expert 1 week in advance


Very important!!!: Read the Presenter’s guide (available on the website)

What if I can’t make my presentation?

• If you want to change / swap your topic, try and find someone else to swap with….

• …if you still can’t find a solution, then get in touch with Maria, Justin or Hanneke as soon as possible (at least 3 weeks before the talk).


Where to find help

• Key papers

• Previous years’ slides

• Human Brain Function Textbook (online)

• SPM course slides

• Cambridge CBU homepage (Rik Henson’s slides)

• Methods Group Experts

• Monday Methods Meetings (4th floor FIL, 12.30)

• SPM email List


MfD Home Resources

http://www.fil.ion.ucl.ac.uk/mfd/page2/page2.html

Experts• Will Penny – Head of Methods

• John Ashburner

• Stephan Kiebel

• Guillaume Flandin

• James Kilner

• Rosalyn Moran

• Carlton Chu

• Andre Marreiros

• Vladimir Litvak

• Zoltan Nagy

• Justin Chumbley

• Hanneke den Ouden

• Maria Joao Rosa


Contact the expert: discuss presentation and other issues (1 week before talk)

Expert will be present in the session

Website

http://www.fil.ion.ucl.ac.uk/mfd/


Where you can find

all the information about MfD 2008:

Programme

Contacts

Presenter’s guide

Resources (Help)

Etc…

Other helpful courses


• Matlab for Cognitive Neuroscience (ICN)– Run by Christian Ruff– http://www.icn.ucl.ac.uk/courses/MATLAB-Tutorials/index.htm– 4.30 pm, Thursday (not every week!)– 17 Queen Square, basement seminar room

• Physics lecture series– Run by FIL physics team– Details will be announced– 12 Queen Square, Seminar room

http://www.icn.ucl.ac.uk/courses/MATLAB-Tutorials/index.htm

Overview for Dummies


Outline

• Getting started with an experiment

• SPM & your (fMRI) data– Preprocessing– Analysis– Connectivity

• Acronyms


Getting started – Cogent

• http://www.vislab.ucl.ac.uk/Cogent/

– present scanner-synchronized visual stimuli, auditory stimuli,

mechanical stimuli, taste and smell stimuli

– monitor key presses

– physiological recordings

– logging stimulus & scan onset times

• Try and get hold of one to modify rather than starting from scratch!

People are more than happy to share scripts around.

• If you need help, talk to Eric Featherstone.


Getting started - Setting up your experiment

If you need…

• special equipment– Peter Aston

– Physics team

• special scanning sequences– Physics team

• They are very happy to help, but contact them in time!


Getting started - scanning decisions to be made

• What are your scanning parameters:

– how many conditions/sessions/blocks

– Interstimulus interval

– Scanning sequence

– Scanning angle

– How much brain coverage do you need

• how many slices

• what slice thickness

– what TR

• Use the physics wiki page: http://cast.fil.ion.ucl.ac.uk/pmwiki/pmwiki.php


http://cast.fil.ion.ucl.ac.uk/pmwiki/pmwiki.php

http://cast.fil.ion.ucl.ac.uk/pmwiki/pmwiki.php

Summary

• Get you script ready & working with the scanner

• Make sure it logs all the data you need for your analysis

• Back up your data from the stimulus PC! You can transfer it via the network after each scanning session…

• Get a scanning buddy if it’s your first scanning study

• Provide the radiographers with tea, biscuits, chocolate etc.


Use the project presentations!

They are there to help you design a project that will get you

data that can actually be analyzed in a meaningful way


Hurrah! I have brain data!

• So what do I do now?

• This is where we get into SPM & preprocessing…

• …and more decision-making!

• All the processing takes a long time, so make sure you have decided in advance, and don’t need to redo your analysis


Statistical Parametric Mapping

• MfD 2008 will focus on the use of SPM8• SPM software has been designed for the analysis of brain imaging

data in fMRI, PET, SPECT, EEG & MEG • It runs in Matlab…just type SPM at the prompt and all will be

revealed.• There are sample data sets available on the SPM website to play

with

Preprocessing Possibilities…

• These steps basically get your imaging data to a state where you can start your analysis

– Realignment & Unwarping

– Segmentation and Normalisation

– Smoothing

Analysis

• Once you have carried out your pre-processing you can specify your design and data

– The design matrix is simply a mathematical description of your experiment

E.g. ‘visual stimulus on = 1’ ‘visual stimulus off = 0’

Analysis

• Our fMRI data is a time series based on the haemodynamic response. The basis functions used in SPM are curves used to ‘describe’ or fit the haemodynamic response in relation to our model



– E.g. ‘visual stimulus on = 1’ ‘visual stimulus off = 0’

Analysis

• Our fMRI data is a time series based on the haemodynamic response. The basis functions used in SPM are curves used to ‘describe’ or fit the haemodynamic response in relation to our model



– E.g. ‘visual stimulus on = 1’ ‘visual stimulus off = 0’

• The HRF is convolved with the design matrix, and we estimate how much variance of the BOLD response our convolved parameters can explain for each voxel, which is expressed in an SPM

Contrasts & inference

• The SPMs are then thresholded to correct for multiple comparisons

• Contrasts allow us to test hypotheses about our data, using t & f tests

• 1st level analysis: activation over scans (within subject)

• 2nd level analysis: activation over subjects

Write up and publish…

Connectivity

• Functional segregation – responses to an input giving a regionally specific effect

• Functional integration – how one region influences another…subdivided into: – Functional connectivity: correlations among brain systems (e.g.

principal component analysis)– Effective connectivity: the influence of one region over another

(e.g. psycho-physiological interactions, or DCM)

Acronyms

• DCM – dynamic causal model• DTI – diffusion tensor imaging • FDR – false discovery rate• FFX – fixed effects analysis• FIR – finite impulse response• FWE – family wise error• FWHM – full width half maximum• GLM – general linear model• GRF – gaussian random field theory• HRF – haemodynamic response

function• ICA – independent component

analysis• ISI – interstimulus interval

• PCA – principal component analysis• PEB – parametric empirical bayes• PPI – psychophysiological interaction• PPM – posterior probability map• ReML – restricted maximum likelihood• RFT– random field theory• RFX – random effects analysis• ROI – region of interest• SOA – stimulus onset asynchrony• SPM – statistical parametric mapping• VBM – voxel-based morphometry

Documents

Introduction / Overview 8th October 2008 Hanneke den Ouden, Justin Chumbley, Maria Joao Rosa Wellcome Trust Centre for Neuroimaging, UCL