Skip to content

Latest commit

 

History

History
72 lines (63 loc) · 3.67 KB

future_additions.rst

File metadata and controls

72 lines (63 loc) · 3.67 KB
Raw

Future updates

Note

Any update ideas can be given here: https://github.com/elliohow/fMRI_ROI_Analysis_Tool/discussions

General

  • Test fRAT with empty parameters for cases where a single metric is set to be reported and not a comparison between multiple parameters - May need to find references to config.parameter_dict/cfg.parameter_dict to see how it is handled when blank
  • Summarise data across an ROI (time series analysis)
  • Add support for custom statistical map functions
  • Add option to only keep essential files when running statistical map functions (such as '_tSNR')
  • To facilitate easier testing of new features during development, add ability to test only ROI analysis, only statistical map creation, or both.
  • Dash barchart code needs an overhaul as it isn't currently viable to easily add new features
    • Long names currently get cutoff
  • Give user ability to find fRAT version
  • Print log of terminal output. This will allow the user to track where fRAT terminated in the case of a crash
  • Save path of last analysed folder so its easier to string together statmap creation and ROI analysis
  • Separation of noise volumes currently returns false positives
  • Add option to use atlases with the 25 or 50 probability threshold
  • (Potential) Add ability to analyse more than fMRI volumes (such as anatomical volumes)
  • (Potential) Add easy way load analysis results data so jupyter notebook can be used to explore the data with graphs/statistics

GUI

  • Separate fRAT into 3 components (main analysis, creation of statmaps such as tSNR, general fMRI tools)
  • Ability to make saveable analysis profiles to quickly swap between settings
    • Need to also add ability to keep toml files up to date if any new options have been added/removed
  • Increase linux compatibility

Analysis

  • Add option to bootstrap confidence intervals
  • Allow registration to more templates (such as http://ccraddock.github.io/cluster_roi/)
  • Allow registration to custom templates and ROI atlases
  • Add ability to skip running a file through analysis if file has already been ran through analysis
    • May be good for combining large datasets or adding new participants to the analysis
    • May only be possible with participant averaged data
    • Currently output folder is deleted when rerunning analysis due to mcf files
    • Need to find a way to deal with files which have now been set to ignore in paramValues.csv
    • Need to also force the analysis to rerun if parameter values have changed
  • Increase clarity of output files
  • fRAT should check paramValues.csv before running anatomical file alignment to see if any subjects can be skipped

Figures

  • 3d representation of the brain grid data
  • Option to use scientific notation for brain figures: nilearn/nilearn#2220
  • Ability to add manual lines to graphs (for example, to represent minimum usable tSNR)
  • Improve customisation options of figures (for example, currently barcharts look too big if fewer parameters selected)
  • Add option to add number of voxels to histogram plots
  • Give one IV barcharts colour
  • Implement mixed-roi scaled brain figures
  • Add option to underlay MNI template under each brain grid image
  • (Potential) Interactive plots

Statistics

  • Add ability to compare if linear mixed model coefficients significantly vary between regions
  • Add figures and testing for linear mixed model assumptions (linearity, normality and homogeneity)
  • Add r^{2} statistics to t-tests

Statmaps

  • Handle the func_preprocessed folder more elegantly, currently it is replaced each time statmaps are created