420 likes | 438 Views
Learn how to use the SPM5 User Interface for spatial preprocessing in fMRI analysis, including realignment, coregistration, segmentation, normalization, and smoothing. Follow detailed steps and tips for each stage.
E N D
Buttons in SPM5 Seán O’Sullivan, ION Alice Jones, Dept of Psychology Methods for Dummies 16th Jan 2008
SPM5 – WTF? • Ladles and jellyspoons,I come before youto stand behind youand tell you somethingI know nothing about.Next Thursday,the day after Friday,there will be a ladies' meetingfor men only.Wear your best clothes,if you haven't any,and if you can come,please stay home. Admission is free,you can pay at the door. We'll give you a seat,so you can sit on the floor. It makes no differencewhere you sit,the kid in the galleryis sure to spit.
Topics • Introducing the SPM5 User Interface • Spatial Pre-processing in fMRI • Model design • 1st and 2nd Level Analysis • Help in SPM5
Preprocessing Analysis Inference
SPM5 User Interface Current List of Jobs
SPM5 User Interface Options available for currently highlighted object Current List of Jobs Current value of / information about highlighted object Save/Load as .mat files or XML (“load-xml”, “savexml”) Info about the meaning of current item
Spatial Pre-processing • Realignment • Coregistration • Segmentation • Normalize • Smoothing
Analysing with SPM5 • Data are available from • http://www.fil.ion.ucl.ac.uk/spm/data/ • Create a new directory for data • Create a subdirectory “jobs” • Open MATLAB • Get into the correct working directory • Type “SPM fmri” • If you’re using SPM for the first time, make sure you “Set Path”, under File in MATLAB. Enter the path to your SPM folder and select the “Add with Subfolders” option
Realignment • Click on “Realign” from drop-down menu
Realignment • Select “New Realign:Estimate and Reslice” • Open “Realign:Estimate and Reslice” option • Highlight Data and select “New Sesson” • Highlight “Session” • Select “Specify Files”
Realignment • Choose all of the functional images in the directory • i.e. images beginning ‘fM000*.img’
Realignment • Save job file as e.g. “realign.mat” in your jobs directory • Press “RUN”
Realignment etc Header files modified with orientation info Mean image for use in coregistration
Coregistration • Click on “Coregister”
Coregistration • Click on “New Coreg:Estimate” • Double-click on “Coreg:Estimate” • Highlight “Reference Image” • select mean fMRI scan meanfM00223_004.img from realignment • Highlight “Source Image” • select structural image sM00223_002.img • SAVE as ‘coreg.mat’ • Press “RUN”
Coregistration • Effects: • SPM implements a coregistration between structural and functional data that maximises mutual information • SPM changes header of source file i.e. sM00223_002.hdr
Coregistration • Useful to check registration of ref and source images at this point • Click “Check Reg” button • Select your source and ref images as before • Navigate images and inspect anatomical correspondence
Segmentation • Click on “Segment”
Segmentation • Highlight Data field • “Specify Files” • select the subject’s registered structural image sM00223_002.img • SAVE as segment.mat • RUN
Segmentation • Effects: • SPM creates grey and white matter images and a bias-field corrected structural image • View with Check Reg • Grey matter image is c1sM00223_002.img • White matter image is c2sM00223_002.img • Check reg against original structural sM00223_002.img Grey matter image • SPM also writes spatial normalisation and inverse spatial normalisation parameters to files in structural directory: • sM00223_002_seg_sn.mat • sM00223_002_seg_inv_sn.mat • THESE CAN BE USED TO NORMALISE FUNCTIONAL DATA Original structural image
Normalize • Click on “Normalize”
Normalize • Select “Normalise:Write” • Allows previously determined warps to be applied to a series of images • Highlight “Data” • Select new “Subject” • Open “Subject” and highlight “Parameter File” • Select sM00223_002_seg_sn.mat from Segmentation step • Highlight “Images to Write” • “Specify Files” • Use filter to select all realigned functional images • Type ^r.* in SPM file selector and click “Filt” • Right-click “Select all” • Done
Normalize • Open “Writing Options” • Click “Voxel sizes”, then “Specifiy Values” • Change values to [3 3 3] • This writes images at a resolution closer to that at which they were acquired • SAVE as “normalise.mat • RUN
Normalize • Effects: • SPM writes spatially normalised files to the functional data directory • Normalised files have the prefix “w”
Smooth • Click on “Smooth”
Smooth • Open “Smooth” • Select “Images to Smooth” • select the spatially normalised files “wrfM00*.img” • Highlight “FWHM” • “Specify Values” • Change [8 8 8] to [6 6 6] • Data will be smoothed by 6mm in each direction • SAVE as smooth.mat • RUN
Smooth • Effects • See right • Normalised functional image above wrfM00223_004.img • Smoothed image below swrf00223_004.img • Note: • SPM5 Manual says “smoothing step is unnecessary if you are only interested in Bayesian analysis of your functional data”
Overview • fM00223_004.img • Realign • rfM00223_004.img • Coregister • Segment • Normalise • wrfM00223_004.img • Smooth • swrfM00223_004.img
First Level Analysis 3 STAGES • Specification of GLM design matrix, fMRI data files and filtering. 2. Estimation of GLM parameters 3. Interrogation of results using contrast vectors to product Statistical Parametric Maps or Posterior Probability Maps.
Model specification, review and estimation box Specify 1st-level Starting 1st level analyses
Building a Design Matrix • fMRI Model Specification • Directory • specify files (ie. where you want the .mat file to be written) • Timing Parameters • Unit for design scans or seconds • Interscan Interval TR (ie. time taken between acquisitions) • Data & Design • Subject/Session • Scans load sw. files • Conditions
Subject/Session Data & Design Load sw. files Add number of conditions required Name of condition Time of onsets (remember scan/seconds) Duration (remember scan/seconds) Also here: Regressors Covariates Masks etc
Estimation • Model parameters can be estimated using classical (Restricted Maximum Likelihood) or Bayesian algorithms • Select Estimate • from the panel on the right and select SPM.mat file you have just created.
Results Define contrasts
Second Level Analyses Second level analyses allow you to make populations inferences from your data. As in first level, you will; Configure design matrix Describe General Linear Model Specify data to be used (.con images) Include other parameters relevant to your study (covariates, global normalisation options, grand mean scaling options, masking and thresholds etc) As for first-level: design and data configuration is followed by ‘ESTIMATE’ and building contrast maps in ‘RESULTS’
HELP! In SPM5 There are a few ways to access help in SPM5 The Help button on the GUI This brings up a helpful display where clicking on a button brings up information about that function.
HELP! In SPM5 • Help can also be obtained from clicking the ? button
HELP! In SPM5 Many options automatically provide a brief explanation of what they might be used for or when to select them
Sources of plagiarism • Alice Grogan, Carolyn McGettigan Buttons in SPM5 • SPM5 Manual - The FIL Methods Group