#This is the FAQ template
# Make sure you "Add to" CategoryFaq
<<FslToolContents>>
= Step by Step Guide =
== Resting state analysis (with dual regression) ==
 * [[InitialProcessing|Initial processing]] (dicom conversion, fslreorient2std, brain extraction for structurals)
 * [ Optional: denoise individual datasets with [[FIX]] or via manual classification and denoising ]
 * Run ''concat ICA'' via the [[melodic|Melodic GUI]] for all sessions/subjects (go through each tab and fill in appropriate information)
     * ''Misc'' tab: defaults are fine
     * ''Data'' tab: select session/subject dataset (if multiple sessions per subject exist then include individual sessions as separate inputs, and do not concatenate the data)
     * ''Pre-Stats'' tab: the defaults here represent our general recommendations
     * ''Registration'' tab: we recommend using a `Main structural image`, with `BBR`, plus `nonlinear` registration to the `Standard space`
     * ''Stats'' tab: select `Multi-session temporal concatenation`
     * ''Post-Stats'' tab: the settings here are unimportant in this instance
 * Check the webpage output, especially the registrations
 * Look at output components and decide what networks are of interest
 * Run the Glm GUI (for higher-level analysis) 
     * Setup an appropriate group design (e.g. unpaired group difference) - the `Wizard` offers some simple/common designs
     * Save the design (need `.mat` and `.con` files later)
 * Run the [[DualRegression/UserGuide|dual_regression]] script, from the directory where the melodic outputs were written (use the GLM design files, `melodic_IC` image, `.filelist` file) 
 * Look at the outputs from dual regression
     * These outputs are from [[randomise/UserGuide|randomise]] and the *corrp* images are the crucial ones to view (these are the multiple comparison corrected p-values, stored as 1-p values)