• CorticalParcellation

top

Cortical Parcellation

The FreeSurfer utilities mris_ca_train and mris_ca_label together implement a technique for automatically assigning a neuroanatomical label to each location on a cortical surface model based on probabilistic information estimated from a manually labeled training set (made using FreeSurfer). This procedure incorporates both geometric information derived from the cortical model, and neuroanatomical convention, as found in the training set. The result is a complete labeling of cortical sulci and gyri.

Background

• Automatically Parcellating the Human Cerebral Cortex, Fischl et al., (2004). Cerebral Cortex, 14:11-22.

• An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest, Desikan et al., (2006). NeuroImage, 31(3):968-80.

• DKT40 classifier atlas: FreeSurfer atlas (.gcs) from 40 of the Mindboggle-101 participants (2012)

Utilities

• mris_sample_parc - samples a volumetric parcellation onto a surface

• mris_label2annot - creates a .annot file from a set of labels and a colortable, for input to mris_ca_train

• mris_ca_train - creates the reference atlas (produces .gcs files from a set of .annot files)

• mris_ca_label - parcellates a cortical surface based on the reference atlas (produces .annot files)

Atlases

• There are three atlases included with Freesurfer:

  • $FREESURFER_HOME/average/?h.curvature.buckner40.filled.desikan_killiany.2010-03-25.gcs - see Desikan et al., (2006).

    • This atlas is also known as the 'Desikan-Killiany' cortical atlas. It is a gyral based atlas: ie, a gyrus was defined as running between the bottoms of two adjacent sulci. That is, a gyrus includes the part visible on the pial view + adjacent banks of the sulci limiting this gyrus.

    • Subject output is named <subjid>/label/?h.aparc.annot

  • $FREESURFER_HOME/average/?h.destrieux.simple.2009-07-29.gcs - see Fischl et al., (2004) and Destrieux et al. (2010).

    • This atlas is also known as the 'Destrieux' cortical atlas. It is based on a parcellation scheme that first divided the cortex into gyral and sulcal regions, the limit between both being given by the curvature value of the surface. A gyrus only includes the cortex visible on the pial view, the hidden cortex (banks of sulci) are marked sulcus.

    • Subject output is named <subjid>/label/?h.aparc.a2009s.annot

    • Atlas changed August 2009. See DestrieuxAtlasChanges.

  • $FREESURFER_HOME/average/?h.DKTatlas40.gcs

    • DKT40 classifier atlas

    • 40 of the 101 mindboggle subjects were used

    • script used to build

• To create your own atlas, see the example in mris_ca_train. See also: SurfaceRegAndTemplates and tksurfer_labeledit.

.annot files

After Freesurfer processes a subject, in the subject's /label directory, there are .annot files containing the parcellation data for each hemishere.

• ?h.aparc.annot files contain the desikan_killiany.gcs parcellation scheme
• ?h.aparc.a2009s.annot files contain the destrieux.simple.2009-07-28.gcs scheme
• ?h.aparc.DKTatlas.annot correspond to the DKTatas40.gcs scheme

The data in the .annot files can be read using the matlab script read_annotation.m (also found in $FREESURFER_HOME/matlab). Example usage:

>> [vertices,label,colortable]=read_annotation('rh.aparc.annot');

Where vertices is just each vertex number. label contains the parcellation label for that vertex, where the label is a colortable number. For instance, if a label is '6553700', then find that number in the colortable.table structure, and its label name is that same index in the colortable.struct_name structure (in this example, '6553700' is 'frontalpole').

Lobe mapping

While some ROIs cross two or more lobar boundaries (i.e. fusiform), others (i.e. cingulate subdivisions) can be considered separately or can be included to follow the 4 lobes. The following is an approximate mapping of individual 'Desikan-Killiany' ROIs (found in ?h.aparc.annot) to the lobes, described in the Appendix of this publication:

Frontal

• Superior Frontal
• Rostral and Caudal Middle Frontal
• Pars Opercularis, Pars Triangularis, and Pars Orbitalis
• Lateral and Medial Orbitofrontal
• Precentral
• Paracentral
• Frontal Pole

Parietal

• Superior Parietal
• Inferior Parietal
• Supramarginal
• Postcentral
• Precuneus

Temporal

• Superior, Middle, and Inferior Temporal
• Banks of the Superior Temporal Sulcus
• Fusiform
• Transverse Temporal
• Entorhinal
• Temporal Pole
• Parahippocampal

Occipital

• Lateral Occipital
• Lingual
• Cuneus
• Pericalcarine

Cingulate (if you want to include in a lobe)

• Rostral Anterior (Frontal)
• Caudal Anterior (Frontal)
• Posterior (Parietal)
• Isthmus (Parietal)

You can use mri_annotation2label to extract the individual labels of an existing cortical parcellation and then mri_mergelabels to fuse those labels together to form a lobe.

mri_annotation2label  --subject subjid --hemi rh --outdir ./location
mri_mergelabels -i label1 -i label2 -o outputlabel

For FS V5.0 and later, you can also run mri_annotation2label with --lobesStrict to get a lobe annotation. If that definition of "lobes" is good for you, then you can run mris_anatomical_stats to get the volume for each lobe.

For earlier versions, you can use mri_annotation2label to break the labels apart, then use mri_mergelabels to combine the individual labels into lobe labels, then use mris_label2annot to create a lobe annotation, then use mris_anatomical_stats.

mri_annotation2label --help will show you more options that are available to you.