Shed

Shed – Automated parcellation of functional activation maps

Description

Shed is designed to perform automated sub-division of a statistical map into regions derived from the topology of the map. The Watershed algorithm is used to identify separable peaks from within contiguous super-threshold clusters. The program is written in Matlab and provides a graphical user interface to select input files and operating parameters. It is specifically intended for use with first- or second-level activations maps from functional MRI studies. The program reads and writes 3D NIFTI format images.

Procedure

Shed works in the following manner:

  • An activation map, and optional underlay anatomical image, are loaded by the user
  • Using statistical and extent thresholds, the activation maps is divided into standard clusters
  • The Watershed algorithm is used to divide clusters into regions with separable peak sub-regions
  • A label image is saved to disk along with a print-out of region statistics
Features
  • Shed is implemented through a GUI with no user intervention (i.e. no hand drawing).
  • Reads and writes 3D NIFTI format images.
  • Implements user-selectable settings for threshold, extent, neighboring voxel definition, and others options
  • Displays the results in the GUI for user inspection.
Requirements
Installation Instructions
  • Download the MATLAB programs (see Download section below).
  • Add the files to your Matlab Path.
  • For Linux or Mac, copy the file shed.fig.LINUX to shed.fig.
  • For Windows, copy the file shed.fig.WINDOWS to shed.fig.
  • Be sure that you have the Tools for NIfTI and ANALYZE Images folder in your Matlab path.
  • Run Matlab and type shed at the prompt to bring up the Shed GUI.
Downloading

The Shed program is available on GitHub: github.com/markymarkymark/Shed.git

Documentation
  • Load an activation map by clicking the Activation Map… button. Select a NIFTI file (extension .nii)
  • Load an optional anatomical image with the same voxel dimensions by clicking on the Struct Image… button.
  • Click the GO button to run the program
Contact

For comments or questions on the Shed program, send e-mail to Mark Elliott (melliott@pennmedicine.upenn.edu)