Bibliography of Segmentation in Neuroimaging

Finn Årup Nielsen
CIMBI at DTU Informatics and NRU Rigshospitalet
Lyngby and Copenhagen, Denmark

  $Revision: 1.75 $
  $Date: 2009/03/11 15:24:52 $

Abstract:

Reference for segmentation in neuroimaging are collected. Both tissue segmentation and parcellation is included.

This structured bibliography is part of a larger collection of bibliographies see http://www.imm.dtu.dk/~fn/bib/Nielsen2001Bib/. The bibliography is written in L^ATEX and BIBTeX and should be available both as HTML and PostScript.

The bibliography is probably far from complete, but new references are added whenever the author finds new material and has the time to add them. You can email the author if corrections are required or you have found some reference that you fell ought to be included: fn@imm.dtu.dk.

Thanks to Jürgen Hänggi, Jonathan Bailleul and Arno Klein who provided information. Much of the information in this bibliography is from the SPM mailing list posted by numerous researchers.

This work is or has been funded by the European Union project MAPAWAMO, the International Neuroimaging Consortium (INC) American Human Brain Project, Danish Research Councils through THOR Center for Neuroinformatics, the Villum Kann Rasmussen Foundation and the Lundbeck Foundation.

List of Tables

General references

A list of references is available from http://neuro-www.mgh.harvard.edu:16080/cma/seg/references.html

Unclassified

[Harris et al., 2001] parcellation of cortex with brain warping and manual atlas. PMID: 8978636, PMID: 9786148, PMID: 11185422

B. Dawant, S.L. Hartmann, J.-P. Thirion, F. Maes, D. Vandermeulen, P. Demaerel, Automatic 3-D segmentation of internal structures of the head in MR images using a combination of similarity and free-form transformations : part I, methodology and validation on normal subjects , IEEE transactions on medical imaging, vol. 18, no. 10, pp. 909-916, October 1999

Inhomogeneity correction

MRI intensity non-uniformity (intensity inhomogeneity) can have a substantial impact on the performance of the segmentation results [Sled et al., 1997a] and the image should be bias field corrected. Table 1 shows a number of the tools in use. [Ashburner and Friston, 1998] describes a combination of tissue classification with inhomogeneity correction.

[Arnold et al., 2001,Schaper et al., 2001] compared six algorithms for inhomogeneity correction (N3, hum, eq, bfc, cma and SPM99). [Boyes et al., 2008] investigated the performance of the N3 program on scans from 3T scanners. Other references in relation to bias field estimation are [Guillemaud and Brady, 1997] and a review [Hou, 2006].

Name	Method and Description	References
cma		Center for Morphometric Analysis, Massachusetts General Hospital
EMS *	Polynomial basis functions. Part of segmentation program	[Van Leemput, 2001, p. 15-19], [Van Leemput et al., 1999a]
eq		[Cohen et al., 2000]
FAST *	``FMRIB's Automated Segmentation Tool''. A segmentation tool including inhomobeneity correction	http://www.fmrib.ox.ac.uk/fsl/fast/index.html
FreeSurfer *	Implemented in the `mri_normalize` program. Can be executed from `csurf` GUI.	[Dale et al., 1999,Fischl et al., 1999a,Fischl et al., 1999b,Fischl and Dale, 2000,Fischl et al., 2001,Busa, 2002] http://surfer.nmr.mgh.harvard.edu/
hum		[Brinkmann et al., 1998]
ITK *	The ``itk::MRIBiasFieldCorrectionFilter'' class in the National Library of Medicine Insight Segmentation and Registration Toolkit (ITK) based on Legendre polynomial	http://www.itk.org-/HTML-/MRIBiasCorrection.htm, [Styner et al., 2000,Styner and Gerig, 1997]
N3 *		[Sled et al., 1998,Sled et al., 1997b,Sled, 1997], http://packages.bic.mni.mcgill.ca/, http://www.bic.mni.mcgill.ca/software/N3/
PABIC *	``PArametric BIas field Correction''. Included in ITK.	[Styner et al., 2000,Styner and Gerig, 1997]
SPM *		[Ashburner and Friston, 2000]
SPM2 *	Available in the functions with prefix `spm_bias_`	[Ashburner, 2002], Early version: ftp://ftp.fil.ion.ucl.ac.uk/spm/flatten
vol_homocor.m *	Program by Gary Glover distributed by Kalina Christoff	http://www-psych.stanford.edu/~ kalina/SPM99/Tools/vol_homocor.html

Stripping

The process of ``stripping'', ``skull-stripping'', ``brain/non-brain segmentation'', ``brain surface extraction'', ``brain extraction'' or ``brain extraction algorithms (BEA)'' removes the skull, scalp and meninges and maintains the ``brain'' which usually includes white and grey matter as well as CSF (at least the ventricular CSF). Table 2 is a list of the tools for this operation. A study showed that McStrip was much slower than BSE and BET, but that it was the most precise [Boesen et al., 2003]. In a comparison BET, 3dIntracranial, HWA and BSE against a manual stripping as gold standard ``BSE tended to perform best'' and ``HWA and BSE were more robust across diagnostic groups'' [Fennema-Notestine et al., 2006]. Another algorithm is described in [Atkins and Mackiewich, 1998].

One study found that in voxel-based morphometry (VBM) using SPM2 brain extraction would profoundly affect the results [Fein et al., 2006].

**Table 2:** Stripping
Name	Impl.	Description	Reference

3dIntracranial *		Brain extraction included in AFNI	[Ward, 1999], http://afni.nimh.nih.gov/afni/doc/help/3dIntracranial.html
BEMA		``Brain extraction meta-algorithm''	[Rex et al., 2004]
BET *		``Brain Extraction Tool'' by Stephen Smith. Conveniently included in FSL, MRIcro and mri3dX	[Smith, 2002,Smith, 2000], http://www.fmrib.ox.ac.uk/fsl/bet/, MRIcro: http://www.psychology.nottingham.ac.uk/staff/cr1/mricro.html
BSE *		``Brain Surface Extraction''part of BrainSuite. Interactive GUI version exists with the X/Motif-based xbse	[Shattuck et al., 2001,Sandor and Leahy, 1997], http://neuroimage.usc.edu/BSE/
McStrip *	IDL, C	``Minneapolis Consensus Strip'' (MCS). Consensus/hybrid based method relying on AIR5.0 and BSE	[Rehm et al., 2004,Rehm et al., 1999], http://www.neurovia.umn.edu/incweb/McStrip_download.html
MIPAV *	Java		[Bazin et al., 2007,Goldszal et al., 1998], http://mipav.cit.nih.gov/
FreeSurfer *		Can be executed from `csurf` GUI.	[Dale et al., 1999,Fischl et al., 1999a,Fischl et al., 1999b,Fischl and Dale, 2000,Fischl et al., 2001,Busa, 2002] http://surfer.nmr.mgh.harvard.edu/
HWA *		Hybrid Watershed algorithm in FreeSurfer	[Segonne et al., 2004]

Brain tissue segmentation

Brain tissue segmentation typically classifies voxels into grey matter, white matter, CSF and ``non-brain''. Some segmentations works with a further ``lesion'' class.

Methods for segmentation

Many papers describe methods for brain tissue segmentation, and just a few are listed in Table 3. Others are [Cocosco et al., 2002,Sun and Wang, 2005].

**Table 3:** Methods for segmentation.
Description	Reference

Use of atlas prior (tissue probability maps in stereotaxic space)	[Kamber et al., 1995]
Input as T1, T2, PD and output as GM, WM, CSF. Selection of a training classes for the segmentation	[Harris et al., 1999]
Input as T1, T2, PD and output as WM, GM, CSF, outliers. With inhomogeneity correction and atlas prior. Gaussian mixture estimated robustly. Lesions detected as outliers. Bias field modeled with polynomials. Markov random field for prior volumes	[Van Leemput et al., 2001,Van Leemput et al., 2000]
K-Nearest Neighbor on data from five types of regular MRI-scans for classification of white matter lesions	[Anbeek et al., 2003]
``Fuzzy inference system'' on 3 different MR images for classifying white matter hyperintensity	[Admiraal-Behloul et al., 2005]
Support vector machine on 4 different MR images for white matter lesion segmentation	[Lao et al., 2006]

A ``ground truth'' makes it possible to evaluate the performance of the segmentation algorithm. [Moretti et al., 2000] took this approach with the use of the BrainWeb labeled brain as ground truth.

Tools

**Table 4:** Tools for segmentation
Name	Input	Output	Description	Reference

BrainSeg				Ali Hojjat
EMS			`Expectation-Maximization Segmentation' implemented as an SPM plugin	[Van Leemput et al., 2001,Van Leemput et al., 1999b,Van Leemput et al., 1999a,Maes et al., 1997,Van Leemput et al., 2000] http://bilbo.esat.kuleuven.ac.be/web-pages/downloads/ems/ems.html
FAST *		GM, WM, CSF, ...	FMRIB's Automated Segmentation Tool, Hidden Markov model with inhomogeneity correction	[Zhang et al., 2001a,Zhang et al., 2000,Zhang et al., 2001b] http://www.fmrib.ox.ac.uk/fsl/fast/
INSECT			GM, WM and CSF segmentation with an artificial neural network with 9-parameter spatial normalization	[Kollokian, 1996,Collins et al., 1994]
IRIS *			Visualization program with manual drawing by Guido Gerig and Sean Ho. One of the versions is called IRIS2000	http://www.cs.unc.edu/~ruffin/iris/
MIDAS (Freeborough)			``Medical Image Display and Analysis Software''. Interactive Unix/X program, with thresholding, region growing and morphological operations	[Freeborough et al., 1997]
SEAL			``Sulcal Extraction and Automated Labelling''	[Goualher et al., 1999]
SEGRAS		WM, GM, CSF, Lesion	Trained artificial neural network used as classifier	Alan Rene Rasmussen, Hvidovre Hospital
SPM *			Segments into GM, WM, CSF and other. Implemented in versions SPM99 and SPM2.	[Ashburner and Friston, 1997,Ashburner and Friston, 2000,Ashburner and Friston, 2003]
SPM5 *			Segmentation with image registration and bias correction	[Ashburner and Friston, 2005]
--			Combined manual/automatic	[Zavaljevski et al., 2000]

**Table 6:** Flattening algorithms
Name	Impl.	Description	Reference

DMflatten			[Balasubramanian et al., 2005,Balasubramanian et al., 2006]
mrFlatMesh	Matlab	Companion to mrGray	A. R. Wade

**Table 7:** Parcellation tools

Name	Description	Reference

ANIMAL	`Automatic Non-linear Image Matching and Anatomical Labeling' Nonlinear warping and labeling by a previous labeled volume	[Collins et al., 1995], http://www.bic.mni.mcgill.ca/users/louis/MNI_ANIMAL_home/readme/ ?
Mindboggle	Automatic labeling based on 20 labeled templates	[Klein et al., 2005,Klein and Hirsch, 2005,Klein and Hirsch, 2001]
pveout	Nonlinear alignment with MRIWarp to 10 different labeled templates	[Svarer et al., 2005]

**Table 9:** Labeled brains. The second column with the `#' heading indicates the number of labels. `*' denotes that the labeled brain is readily available on the Internet. Entries above the line is digitized and paper atlasses are below the line.

Name	#	Description	Reference

AAL		See Tzourio-Mazoyer
Brodmann *	41+1	Brodmann areas. Non-space filling, non-probabilistic.	Van Essen, Drury. Included in MRIcro as `brodmann.hdr`/`brodmann.img`(`.gz`)
CBA	``almost 400''	Atlas incorporated in a commercial program. ``Greitz atlas''. Brodmann areas, gyri, central structures	[Applied Medical Imaging, 1994,Seitz et al., 1990,Greitz et al., 1991,Thurfjell et al., 1994,Thurfjell et al., 1995,Bohm et al., 1986,Bohm et al., 1989,Bohm et al., 1991,Bohm et al., 1985]
Cerefy		Commercial digitized versions of the Talairach and Schaltenbrand atlases and Windows/MacIntosh program.	[Nowinski et al., 2001,Nowinski et al., 1997,Nowinski et al., 1995b,Nowinski et al., 1995a]
Hammers 2002	2 $\times$ 19 + 3(?), 43	Segmentation of MNI single subject Non-probabilistic, space-filling, Non-hierarchical.	[Hammers et al., 2002]
IBSR ``18''	43	``18 Scans: T1-weighted MR Image data with expert segmentations of 43 individual structures''	http://www.cma.mgh.harvard.edu/ibsr/data.html
ICBM label *(?)	58(?)	ICBM Single subject MRI anatomical template. Distributed in Minc format (`ICBM_labels.mnc` and `ICBM_1.0mm_label.mnc`). Almost space-filling. Non-hierarchical.	http://www.loni.ucla.edu/NCRR/Software/ICBM_Template.html, Label names: http://www.loni.ucla.edu/NCRR/Software/ICBM_Template/Templat_Labels.htm
ICBM Kabani	90/91(?)	Parcellation of MNI single subject in accordance with NeuroNames	[Kabani et al., 1998]
`Iowa' (frontal)	11	Parcellations of the frontal cortex by two human raters	[Crespo-Facorro et al., 1999]
`Iowa' (temporal)	16	Parcellation of the temporal neocortex	[Kim et al., 2000]
`Iowa' (cerebral cortex)	41		[Crespo-Facorro et al., 2000]
Mindboggle *	?	Based on 10-20 subjects. Previously `The Whole Brain Atlas' transformed to MNI-space	[Klein and Hirsch, 2005,Klein and Hirsch, 2001,Klein and Hirsch, 2002,Kikinis et al., 1996], http://www.binarybottle.com/mindboggle.html
MNI SPAM	91	Probabilistic volumes in MNI-space	[Evans et al., 1996,Collins et al., 1999]
Jerne, ``Volumes of Interest'') *	100+	MNI-space, probabilistic, space-filled, hierarchical. Approximate volumes based on labeling in the BrainMap database.	[Nielsen and Hansen, 2002], http://hendrix.imm.dtu.dk/services/jerne/ninf/voi.html
PMaps *	18	Probability map of selected cytoarchitectonic areas from Jülich: 1, 3a, 3b, 4a, 4p, 6, 17 (V1), 18 (V2), 41, 44, 45	[Eickhoff et al., 2005c,Eickhoff, 2005,Geyer et al., 1996,Amunts et al., 1999,Amunts et al., 2000,Morosan et al., 2001,Rademacher et al., 2001,Amunts et al., 1998,Morosan et al., 1996,Eickhoff et al., 2005b,Eickhoff et al., 2005a], http://www.fz-juelich.de/ime/spm_anatomy_toolbox, http://www.fz-juelich.de/ime/ProbabilityMaps_eng.html
Svarer *	$2\times 17+1$	10 subjects labeled in native space	[Svarer et al., 2005,Svarer et al., 2002], http://nru.dk/software/
Talairach Daemon *		A program that contains two brain templates: A digitized Talairach atlas and MNI	[Lancaster et al., 2000b,Lancaster et al., 1997a,Lancaster et al., 1997b,Lancaster et al., 2000a]. The labeling is used by the WFU PickAtlas program.
Tzourio-Mazoyer *	2 $\times$ 45(?), 116	MNI-space, non-probabilistic, non-space filled, semi-hierarchical. This is sometimes referred to as ``automated anatomical labeling'' or ``AAL''. The labeled volume is distributed with MRIcro has 116 different labels	[Tzourio-Mazoyer et al., 2002a,Tzourio-Mazoyer et al., 2002b], http://www.cyceron.fr/freeware/. The labeled volume is distributed with MRIcro as aal.hdr/aal.img(.gz) and aal.txt
VOXEL-MAN	?	Commercial program with atlas	[Höhne et al., 1992,Höhne, 1997,Höhne, 2001]

Brodmann		Book with simple drawings of cytoarchitectonic areas.	[Brodmann, 1994]
			Duvernoy 1992
Mai		Book with labeled brain in stereotaxic space	[Mai et al., 1997]
Ono		Book that describe sulcal variability	[Ono et al., 1990]
Schaltenbrand		Book	[Schaltenbrand and Wahren, 1977]
			Szikla et al. 1977
Talairach	`Many'	Book with a labeled brain in stereotaxic space	[Talairach and Tournoux, 1988,Talairach and Szikla, 1967]

Bibliography of Segmentation in Neuroimaging

Abstract:

Stripping

Brain tissue segmentation

Flattening

Parcellation

Labeled brains