NOT currently maintained. Use at own risk.

fslfeatsetup

Python functions to create an FSL FEAT configuration file (.fsf)

Can be run from python command line/REPL or incorporated into a script

Alpha version available on PyPI

pip install fslfeatsetup

Feel free to contribute

DOCUMENTATION generated by pydoc-markdown

On Python Package Index

Overview

The .fsf file is represented by the class FeatSettings, which is constructed with the analysis level and analysis type options.

Each panel of the FEAT GUI is represented by a separate class, taking the initial FeatSettings object as its argument. Each of these objects has a Configure() function taking keyword arguments specifying the options available in that panel of the GUI. These are the MiscOptions, DataOptions, PreStatsOptions, RegOptions, StatsOptions, and PostStatsOptions classes.

For example the checkboxes and inputs on the Misc Options GUI panel correspond to the key-word arguments in the MiscOptions.Configure() function.

MiscOptions.Configure(brainThreshold=10, noiseLevel=0.66, temporalSmoothness=0.05, zThreshold=5.3, cleanupFirstLevel=False, overwriteOriginalPostStats = False, estimateNoiseFromData=False)

(Some of these options pertain only to higher-level analyses)

Dropdown lists are represented by enum-like classes:

PreStatsOptions.Configure(st = FeatSliceTiming.REGULAR_UP)

PreStatsOptions.Configure(st = FeatSliceTiming.TIMING_FILE, sliceTimingFile = "path/to/file")

First-level analysis example

    from fslfeatsetup.FSF import *
    from fslfeatsetup.EVs import *
    from fslfeatsetup.FSFLabels import *

    SubjectFMRI = [ ... ] 
    SubjectStructurals = [ ... ]

    for i in range(len(SubjectFMRI)):
           # initialize the FeatSettings object
           FSF = FeatSettings(FeatLevel.FIRST_LEVEL, FeatAnalysis.FULL_ANALYSIS)
           
           # Configure the Data options
           Data = DataOptions(FSF)
           # The only required inputs are the output FEAT directory, and the list of 
           # FMRI files (or lower-level feats, see Higher Level Analysis example
           Data.Configure("path/to/output/subject_i",[SubjectFMRI[i]])

           # Configure the Miscellaneous options
           Misc = MiscOptions(FSF)
           # There are NO required inputs. Using the defaults specified in my FSL installation. 
           # If fslfeatsetup needs an option that is not in the defaults, it will let you know 
           Misc.Configure()

           # Configure Registration options
           Reg = RegOptions(FSF)
           
           # I can specify a standard to use, or I can go with the default 2mm MNI152, as I am here
           Reg.ConfigureStandardSpace()
           # The only required argument 
           Reg.ConfigureMainStructural([SubjectStructurals[i]])
           
           # If I don't want to use expanded functional data, I simply don't configure it
           # Reg.ConfigureExpandedFunctional([ this would be a list of your expanded functional images ])
           
           
           # Configure Pre-Stats options
           PreStats = PreStatsOptions(FSF)
           # The library has built-in enum-like structures that hardcode the FEAT options
           PreStats.Configure(sliceTiming=FeatSliceTiming.TIMING_FILE,
                                 sliceTimingFile="path/to/slice/timing/file",
                                 bet=True)
                                 
           # Configure Stats options
           Stats = StatsOptions(FSF)
           # using all defaults, so I don't need to specify keyword arguments
           Stats.Configure()
           # Add EVs from custom 3 column text formats. 
           # Note that ONLY the 3-column text file format is currently supported
           # specify the parameters of the convolution function, or use defaults
           Stats.AddFirstLevelEV("myEV1","path/to/my/EV1",Gamma(phase=0, stdev=3, lag=6))
           Stats.AddFirstLevelEV("myEV2","path/to/my/EV2",Gamma())
           Stats.AddFirstLevelEV("myEV3","path/to/my/EV3",Gamma())

           # orthogonalize
           # The argument is a pythonic matrix (list of lists)
           # the size of this matrix will be one larger than the number of EVs
           Stats.OrthogonalizeEVs([ [ 0 for x in range(4)] for y in range(4)])

           # Configure Post-Stats options
           PostStats = PostStatsOptions(FSF)
           # using all defaults except for min and max Z-threshold for rendering
           PostStats.Configure(zmin = 2, zmax= 8)

           # write to .fsf file
           FSF.write("path/to/subject_i/fsf")

Higher-level analysis steps

todo

Changelog

PyPI version	Description
0.3.4	fixes related to the Issues page. No methods or classes are different
0.2.8	Generated an .fsf accepted by FEAT. There are a bunch of hard-coded defaults that still need to be fixed
0.2.5	Fixed gammadelay being blank and FSLDIR not found
0.2.3	Patch: quotations for custom EV files
0.2.1	Patched a silly bug
0.2.0	first "complete" version, with post-stats options and auto generated comments. still very much a work in progress
0.1.2	first stable version - package still a work in progress