=== Start to Finish: complete analysis === Comments about this course - this file: start2finish.1day.txt (under afni_handouts) - discuss where are we starting from - very few slides - mostly hands-on (ask questions of any instructor) - time is short, must be light on details - try to listen and understand, not to remember details - data, handouts, tutorial are on web site (http://afni.nimh.nih.gov) Where are we starting from? - e.g. new to some lab, inherit experiment and data - somewhat familiar with FMRI - new to Unix, tcsh (t-shell) and AFNI Unix: the type of operating system on the computers (mac/linux) tcsh: command-line interpreter in terminal window AFNI: FMRI software package Main overview of 'Start to Finish' Course 1. discuss experiment and data from notes in afni16 handout 2. discuss data used in analysis - getting anat/EPI into AFNI format - creating stimulus timing files 3. overview of analysis - goal: run group analysis on single subject response magnitudes - single subject analysis: common pre-processing plus regression - what is single subject regression? - group analysis: 3dANOVA2 or 3dMEMA (examples) - what is a t-test, for example 4. how to actually analyze subject FT 5. analyze data and inspect results: single subject - the bulk of this class 6. analyze data and inspect results: group 1. Discuss experiment and data from notes in afni16 handout - under home directory/afni_handouts, open afni16_start_to_finish.pdf - review stimuli, timing and data on slides 2-4 - 2 stimulus classes: auditory-reliable, visual-reliable - 3 EPI runs of randomized 20-second stimulus blocks - acquired data, TRs to remove, timing - experiments well designed before acquiring data - good habits for design is for another time 2. Discuss data used in the analysis - get data into AFNI format (from DICOM images or NIfTI datasets) - to3d/Dimon - groups should already know how (if not, consult us) - HANDS-ON: note files in question (anat, EPI, stim timing) - if on mac, click big "X" in dock to start X11 - open terminal window - mac: Applications -> Terminal - linux: click little window icon - type 'ls' to list the contents of the home directory CMD: cd AFNI_data6/FT_analysis CMD: cd FT - HEAD/BRIK, #TRs, command to create, stim timing CMD: afni - look at anat, look at EPI 3. Overview of analysis - goal: find locations in the brain that respond to our stimuli, also locations that show a big difference between them a. each subject: get volumes representing magnitude of response b. group analysis: measure consistency/significance across group - quick overview of single subject analysis - pre-processing remove pre-steady-state TRs, then ... time shift, register, blur, scale - regression: what actually happens? - tell 3dD how to model the data (polort, motion, stimuli) - it computes best fit of the model to the data - output is magnitude of response (per regressor) and significance level - run group analysis with those results - what is a t-test? average magnitude / stdev - how consistently different from 0 are the subject betas? - have 3dANOVA2 and 3dMEMA scripts for this data 4. How to analyze subject FT - write single subject processing script - why are scripts useful (record, subjects, repeatability) - briefly look at AFNI_data6/FT_analysis/s12.proc.FT.align - can produce that with afni_proc.py (recommended) - can produce AP command with uber_subject.py (or uber_proc.py) - main outputs of processing script: - anat, betas and statistics: aligned and in standard space - blur estimates (to help with multiple comparison correction) - much to look at to evaluate processing - see tutorial (AFNI_data6/FT_analysis/tutorial) 5. Analyze subject FT (the bulk of the class) - run uber_subject.py from FT_analysis directory - is still a work in progress - create AP command and peruse, it creates the proc script - create proc script and peruse, it does the work - execute proc script - discuss overview of script while it executes - it is executed under a new subject_results directory - proc script copies inputs into FT.results and processes there - review processing of data along with script - run afni from results directory - review statistical results - focus on threshold and coloring in afni GUI - cluster results, for practice 6. Analyze and inspect group results - see directory AFNI_data6/group_results - run ANOVA script and view results