GroupTransform

GroupTransform prepares functional images from various subjects for group analysis. Any combination of the following steps can be performed in any order: cross-registration, reslicing, normalization, and smoothing. For more information, see the help documents for the individual steps.

Invocation

java  GroupTransform [sub01 sub02 ...]

GroupTransform window

This window is separated into different sections which can be accessed at any time by clicking on the tab labeled for each section.
What really happens in the cross-register step.

Information to be entered

• Functional images
  
  • List of subjects
    Subjects to process. Subject identifiers should be separated by spaces, and each identifier should correspond to a directory in the subject input directory. The subject list can be provided on the command line as arguments to the java invocation.
  • Subject input directory
    Directory containing subject directories, each of which should contain a directory with functional images or a filelist of functional images.
  • Subject output directory
    Directory where the output files and directories will be created. This directory must be writable. Must be specified only if different from the subject input directory.
  • Directory with functional images
    Text file with filenames of functional images to be processed, or the name of a directory with such images. The images can be compressed or in a read-only partition, as they are not modified in any way (unless Normalize is the first step in the sequence; Normalize will try to uncompress any compressed images it finds). This file or directory must be within each individual subject's directory, which in turn is within the subject input directory.
  • Functional slice with AC/PC
    Slice number of the slice (in the functional image) containing the Anterior Commisure, where slice 1 is the top slice in the volume. The Anterior Commisure can be outside the scope of the slices for the functional image.
• Output subdirectories...
  In this window you can specify the names of output subdirectories (relative to each subject's output directory) that are created by Reslice, Normalize, and Smooth.
• Compress input images when finished
  Check this box if you want input images to be compressed (gzipped) after processing so as to save disk space.

---

• Structural images
  
  • Directory with structural images
    Directory that contains the structural images that are to be aligned to the standard reference image.
  • Subject used for std. ref. image
    Subject to use as standard reference image for registration. This subject's structural image must be in the structs directory, and all the other subjects' structural image dimensions must match this one.
  • Suffix for stripped structurals
    Suffix after the subject identifier used in all the subjects' stripped structural images.
  • Structural slice with AC/PC
    Slice number of the slice (in the structural image) containing the Anterior Commisure, where slice 1 is the top slice in the volume. If the Anterior Commisure is not within the scope of the slices for the structural image, another reference slice (one that is within scope) should be chosen. These parameters will be used to align the functional and the structural images along the Z dimension.
  • Structural subdirectories...
    In this window you can specify the names of directories inside the structs directory where CrossRegister will create .air files and resliced and shrunken images.
  • View CrossRegister results
    Check this box if you want to view the standard reference image and the original and resliced images for each subject after cross-registration is complete.

---

• Parameters
  
  • Background Thresholds
    The thresholds for the standard reference image and for other structural images are passed to CrossRegister, which in turn passes them to alignlinear when aligning other structural images to the standard reference image in creating .air files to be used by Reslice. The threshold for functional images is passed to Normalize.
  • Global image mean for normalization
    Target mean value to use when normalizing images.
  • Smoothing parameter
    Smoothing parameter, full-width half-max in millimeters.
  • Parallelism
    Number of subjects (or number of images for Smooth) to process simultaneously. A parallelism setting of 1 will process all subjects sequentially. The parallelism level chosen should not exceed the number of processors available, otherwise performance will suffer. If the parallelism level exceeds the number of subjects being processed, the number of subjects will be used as the parallelism level.

  ---

  • Processing sequence
    This section allows any combination of steps to be performed in any sequence.
    • Choices and Sequence lists
      Items can be added to the Sequence list by double-clicking on an individual item or by highlighting an item and then clicking the Insert button. You can select or deselect multiple items by option-clicking on each item. Similarly for deleting items from the Sequence list. The items in the Sequence list will be run in order when the Run button is clicked. During the run, the step being performed will be highlighted in the Sequence list.
    • Show
      The Show button will show the windows for each step in the Sequence. These individual windows can be closed at any time without any adverse effects. Note: Parameters in the GroupTransform window override parameters in the individual windows when the Run button in the GroupTransform window is clicked.
    • Check
      The Check button will perform preliminary checks and will print out the number of input images, the location of the reference image, and the location of the structs directory. The preliminary checks will also be performed automatically upon clicking the Run button.

  ---

  Note: functional images for all subjects are assumed to have the same dimensions and the same X and Y voxel sizes; if a given subject's X or Y dimensions differ from the rest, an error will occur. Functional images can have Z voxel sizes that are different from those of other subjects; the resulting functionals (after reslicing) will have Z voxel sizes equal to those of the standard reference image. The dimensions and X and Y voxel sizes of functional images after reslicing will be the same as those prior to reslicing.

  ---

  • Menubar
  • Toolbar
  • Output window

  ---

  Last updated Mon Jul 31 10:12:34 EDT 2000