Neuroimaging Data Processing/Processing/Steps/Surface Extraction

From Wikibooks, open books for an open world
Jump to navigation Jump to search
Neuroimaging Data Processing/Processing/Steps
Smoothing Surface Extraction

Concept[edit | edit source]

The topographic, columnar and laminar organization of the cerebral cortex cannot readily be deduced from 3D volumes or slice representations of the brain. Cortical surface maps provide an alternative approach to represent the cortex and its intrinsic two-dimensional structure. Usually a T1-weighted, anatomical 3D- MRI volume is used to produce an individual surface map. The anatomical surface can be co-registered to a surface atlas Coregistration and Normalization and functional images can be transformed and displayed on the anatomical surface.

Different surface representations are available like pial surface, gray matter - white matter boundary, inflated surfaces or spheres. The advantage of inflated surfaces is that information buried in the sulci becomes visible and it gets clearer which regions are actually adjacent within the tissue and which only get close to each other because of cortical folding. Spheres are designed to make a surface-based coordinate system available.

Surface extraction routines comprise a set of different steps which are described in the following. A complete routine was first described by Fisher and Dale [1] [2] and is implemented in FreeSurfer[3], which is probably the most prominent software for surface extraction.

NOTE: Some of these steps can also be used in volume-based analyses. For instance tissue segmentation might be used in order to perform normalization using tissue probability maps (e.g. done in SPM) or before voxel-based morphometry analysis. Also skullstripping is sometimes included in the preprocessing stream without subsequent surface extraction. AND again the different steps should undergo visual quality control since mistakes in the surface extraction will carry on to the whole analysis.

Isosurfaces[edit | edit source]

The surface extraction described here is different from computing isosurfaces from a volume. The latter much more simply extracts one layer of the volume but does not take into account tissue boundaries, cannot be inflated, etc. Isosurfaces can be helpful for visualizations but do not help to delineate surface topography.

Surface extraction routine[edit | edit source]

Correcting for tissue inhomogeneities[edit | edit source]

As for volume based approaches it is important to correct the anatomical images for magnetic field inhomogeneities. This is described in more detail in Field map / bias field correction

Skullstripping[edit | edit source]

Removal of non-brain tissue, see Skull Stripping

Tissue segmentation[edit | edit source]

Classification of the brain tissue into different tissue classes, usually gray matter (GM), white matter (WM) and cerebrospinal fluid (CSF). See Tissue Segmentation

Cutting Planes[edit | edit source]

In order to gain a representation of each hemisphere and to remove subcortical structures two cutting planes are established. One of them is a saggital cut along the corpus callosum, the other one a horizontal cut through the pons.

Constructing the surface[edit | edit source]

Ensuring that there are no topological defects such as handles (two separate parts of the surface are connected by a tunnel) or patches (holes in the surface) but the mesh achieves a spherical topology. Filling interior holes using connected-component analysis.

Tessellating a triangular mesh over the gray-white boundary of each hemispheric volume and deforming the mesh to produce a smooth representation of the gray-white interface and pial surface Ensuring that there are no topological defects such as handles (two separate parts of the surface are connected by a tunnel) or patches (holes in the surface) but the mesh achieves a spherical topology.

Smoothing[edit | edit source]

For cortical surfaces spatial smoothing is only carried out after the surface extraction. This helps to minimize the blurring of fmri signal across sulci or gyri.

Implementation[edit | edit source]

AFNI[edit | edit source]

SUMA (SUrface Mapping with Afni) is a program for surface based functional imaging analysis in the AFNI framework. It allows viewing three-dimensional cortical surface models, and mapping volumetric data onto them. With SUMA, AFNI can simultaneously and in real-time render Functional Imaging data in 4 modes: Slice, Graph (time series), Volume and Surface with direct links between them. Find out about SUMA here: [4][5] or simply try it (given you have installed AFNI, which contains SUMA sourcecode as well) by entering the AFNI environment and typing

suma

In afni_proc.py the block surf can be included. Important options to specify are the anatomical volume which should be aligned with the surface and the surface specification file(s) (one or both hemispheres).

-do_block surf 
-surf_anat ANAT_DSET
-surf_spec spec1 [spec2]

There are also a couple of AFNI functions that can be used to carry out different single steps of the above routine, see the respective chapters.

IsoSurface [6] is a programm for isosurface extraction from a volume. (see above)

Other helpful functions in handling volumes and surfaces may be 3dVol2Surf, ConvertSurface, CompareSurfaces, Surf2VolCoord

FSL[edit | edit source]

SPM[edit | edit source]

References[edit | edit source]

Huettel, S. A., Song, A.W., & McCarthy, G. (2008). Functional Magnetic Resonance Imaging (2nd edition). Sinauer Associates, Inc: Sunderland, Massachusetts, USA.

  1. Anders M. Dale, Bruce Fischl, Martin I. Sereno, Cortical Surface-Based Analysis: I. Segmentation and Surface Reconstruction, NeuroImage, Volume 9, Issue 2, February 1999, Pages 179-194, ISSN 1053-8119, http://dx.doi.org/10.1006/nimg.1998.0395.
  2. Bruce Fischl, Martin I. Sereno, Anders M. Dale, Cortical Surface-Based Analysis: II: Inflation, Flattening, and a Surface-Based Coordinate System, NeuroImage, Volume 9, Issue 2, February 1999, Pages 195-207, ISSN 1053-8119, http://dx.doi.org/10.1006/nimg.1998.0396.
  3. http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial
  4. http://afni.nimh.nih.gov/afni/suma
  5. http://afni.nimh.nih.gov/pub/dist/doc/program_help/suma.html
  6. http://afni.nimh.nih.gov/pub/dist/doc/program_help/IsoSurface.html