Analysis (Subcortical Structure Segmentation)

What is the Analysis tab?

The Analysis tab brings atlas-based subcortical structure labels (e.g. globus pallidus, putamen, subthalamic nucleus) into your own GRE/QSM space, using non-linear registration powered by ANTs.

Warning

Before using this tab, ‘ANTS_HOME’ must be set up (either in SpecifyToolboxesDirectory.m or via the Manage Dependency tool in the Utility tab) and the atlas of your choice must be downloaded. See Subcortical structure segmentation in SEPIA for detailed setup instructions.

Structure of the application

Unlike the other standalones, the Analysis tab does not share the universal I/O panel or Start button - it consists of a single panel:

• Analysis panel, which lets you pick a segmentation atlas; each atlas has its own dedicated input/output fields, options and Start button.

Analysis panel

• Method

  Select one of the three supported atlases:

  Method	Description
  Segmentation - CIT168 Reinf. learn. atlas	See CIT168 Reinforcement learning atlas
  Segmentation - MuSus100 atlas	See Multi-modal-fused magnetic Susceptibility (MuSus-100) atlas
  Segmentation - AHEAD	See Amsterdam Ultra-high field adult lifespan database (AHEAD) atlas

  Each atlas panel accepts input in one of two ways:

  Input Option 1: Run non-linear registration

  • Select a 3D/4D GRE magnitude NIfTI file

  • Select a GRE mask NIfTI file

  • Select a T1w NIfTI file

  • Select a T1w mask NIfTI file

  • (Optional) Select a Chimap NIfTI file

  This runs the full registration pipeline (GRE-to-T1w rigid-body, then T1w-to-atlas non-linear registration) to bring the atlas labels into your GRE space.

  Input Option 2: Provide transformation matrices

  • Select a Chimap in native space NIfTI file

  • Select a GRE-to-T1w rigid-body transformation

  • Select a T1w-to-Atlas affine transformation

  • Select a T1w-to-Atlas Inverse Wrap NIfTI file

  Use this option if you already have the transformation matrices from a previous ANTs run (e.g. from a prior segmentation using the same subject).

  Note

  If a GRE mask is provided (Option 1), the panel uses Option 1; otherwise it falls back to Option 2’s fields.

  • Output directory

    Directory where the segmentation labels and (optionally) intermediate files will be saved. Defaults to the current working directory if left empty.

  • Correct bias field on input images

    If enabled, applies N4 bias field correction to the input images before registration.

  • Automatic contrast matching

    Note

    Available for the ‘MuSus100’ and ‘AHEAD’ atlases only.

    If enabled, matches the hybrid image contrast to the atlas template before registration.

  • Downsample AHEAD atlas, resolution (mm)

    Note

    Available for the ‘AHEAD’ atlas only.

    If enabled, downsamples the (high-resolution) AHEAD atlas to the specified isotropic resolution before applying it, which can substantially speed up registration at the cost of some label precision.

  • Accelerate using label mask

    Warning

    Use with caution! The result is likely different from whole-brain registration.

    If enabled, speeds up registration by only considering the region around the atlas labels rather than the whole brain.

  • Save intermediate files

    If enabled, keeps all the intermediate files generated during registration (e.g. the individual ANTs transformation steps) rather than removing them once segmentation is complete.

  • Start

    Runs the registration/segmentation pipeline for the selected atlas using the settings above.