-
Stephen Smith authoredStephen Smith authored
index.html 26.14 KiB
<!-- {{{ start -->
<HTML><HEAD>
<TITLE>SIENA - Analysis of Structural Brain MRI Data</TITLE>
</HEAD><BODY BACKGROUND="../images/fsl-bg.jpg">
<hr><TABLE BORDER=0 WIDTH="100%"><TR>
<TD ALIGN=CENTER><H1>SIENA Structural Brain Change Analysis</H1>
SIENA - Structural Image Evaluation, using Normalisation, of Atrophy - Version 2.4<br><br>
<a href="#intro">intro</a> - <a href="#tools">tools used</a> - <a href="#siena">SIENA</a> - <a
href="#sienax">SIENAX</a> - <a href="#sienar">voxelwise SIENA statistics</a>
<TD ALIGN=RIGHT><a href="../index.html"><IMG BORDER=0 SRC="../images/fsl-logo.jpg"></a>
</TR></TABLE>
<!-- }}} -->
<!-- {{{ Introduction -->
<a name="intro"></a><p><hr><H2>Introduction</H2>
<p>SIENA is a package for both single-time-point ("cross-sectional")
and two-time-point ("longitudinal") analysis of brain change, in
particular, the estimation of atrophy (volumetric loss of brain
tissue). SIENA has been used in many clinical studies.
<p><b>siena</b> estimates percentage brain volume change (PBVC)
betweem two input images, taken of the same subject, at different
points in time. It calls a series of FSL programs to strip the
non-brain tissue from the two images, register the two brains (under
the constraint that the skulls are used to hold the scaling constant
during the registration) and analyse the brain change between the two
time points. It is also possible to project the voxelwise atrophy
measures into standard space in a way that allows for multi-subject
voxelwise statistical testing.
<p><b>sienax</b> estimages total brain tissue volume, from a single
image, normalised for skull size. It calls a series of FSL programs:
It first strips non-brain tissue, and then uses the brain and skull
images to estimate the scaling between the subject's image and
standard space. It then runs tissue segmentation to estimate the
volume of brain tissue, and multiplies this by the estimated scaling
factor, to reduce head-size-related variability between subjects.
<p>For more detail on SIENA and technical reports, see the <A
HREF="http://www.fmrib.ox.ac.uk/analysis/research/siena/">SIENA web
page</A>.
<p>If you use SIENA in your research, please make sure that you
reference the following articles. You may alternatively wish to use
the brief descriptive methods text and expanded list of references
given below.
<!-- {{{ Referencing SIENA (minimal version) -->
<hr><b>Referencing SIENA (minimal version)</b>
<p>Two-timepoint percentage brain volume change was estimated with SIENA [Smith 2002], part of FSL [Smith 2004].
<br> or
<br>Brain tissue volume, normalised for subject head size, was estimated with SIENAX [Smith 2002], part of FSL [Smith 2004].
<font size=-1><em>
<p>[Smith 2002] S.M. Smith, Y. Zhang, M. Jenkinson, J. Chen, P.M. Matthews, A. Federico, and N. De Stefano.
<BR> Accurate, robust and automated longitudinal and cross-sectional brain change analysis.
<BR> NeuroImage, 17(1):479-489, 2002.
<p>[Smith 2004] S.M. Smith, M. Jenkinson, M.W. Woolrich, C.F. Beckmann, T.E.J. Behrens, H. Johansen-Berg, P.R. Bannister, M. De Luca, I. Drobnjak, D.E. Flitney, R. Niazy, J. Saunders, J. Vickers, Y. Zhang, N. De Stefano, J.M. Brady, and P.M. Matthews.
<BR> Advances in functional and structural MR image analysis and implementation as FSL.
<BR> NeuroImage, 23(S1):208-219, 2004.
</em></font>
<!-- }}} -->
<!-- {{{ Referencing SIENA (more detailed text and references) -->
<hr><b>Referencing SIENA (more detailed text and references)</b>
<p><b>SIENA</b>. Two-timepoint percentage brain volume change was
estimated with SIENA [Smith 2001, Smith 2002], part of FSL [Smith
2004]. SIENA starts by extracting brain and skull images from the
two-timepoint whole-head input data [Smith 2002b]. The two brain
images are then aligned to each other [Jenkinson 2001, Jenkinson 2002]
(using the skull images to constrain the registration scaling); both
brain images are resampled into the space halfway between the
two. Next, tissue-type segmentation is carried out [Zhang 2001] in
order to find brain/non-brain edge points, and then perpendicular edge
displacement (between the two timepoints) is estimated at these edge
points. Finally, the mean edge displacement is converted into a
(global) estimate of percentage brain volume change between the two
timepoints.
<p><b>SIENAX</b>. Brain tissue volume, normalised for subject head
size, was estimated with SIENAX [Smith 2001, Smith 2002], part of FSL
[Smith 2004]. SIENAX starts by extracting brain and skull images from
the single whole-head input data [Smith 2002b]. The brain image is
then affine-registered to MNI152 space [Jenkinson 2001, Jenkinson
2002] (using the skull image to determine the registration scaling);
this is primarily in order to obtain the volumetric scaling factor, to
be used as a normalisation for head size. Next, tissue-type
segmentation with partial volume estimation is carried out [Zhang
2001] in order to calculate total volume of brain tissue (including
separate estimates of volumes of grey matter, white matter, peripheral
grey matter and ventricular CSF).
<p><b>Voxelwise multi-subject SIENA statistics</b>. First, SIENA was
run separately for each subject. Next, for each subject, the edge
displacement image (encoding, at brain/non-brain edge points, the
outwards or inwards edge change between the two timepoints) was
dilated, transformed into MNI152 space, and masked by a standard
MNI152-space brain edge image. In this way the edge displacement
values were warped onto the standard brain edge. Next, the resulting
images from all subjects were fed into voxelwise statistical analysis
to test for.....
<font size=-1><em>
<P>[Smith 2001] S.M. Smith, N. De Stefano, M. Jenkinson, and P.M. Matthews.
<BR> Normalised accurate measurement of longitudinal brain change.
<BR> Journal of Computer Assisted Tomography, 25(3):466-475, May/June 2001.
<P>[Smith 2002] S.M. Smith, Y. Zhang, M. Jenkinson, J. Chen, P.M. Matthews, A. Federico, and N. De Stefano.
<BR> Accurate, robust and automated longitudinal and cross-sectional brain change analysis.
<BR> NeuroImage, 17(1):479-489, 2002.
<P>[Smith 2004] S.M. Smith, M. Jenkinson, M.W. Woolrich, C.F. Beckmann, T.E.J. Behrens, H. Johansen-Berg, P.R. Bannister, M. De Luca, I. Drobnjak, D.E. Flitney, R. Niazy, J. Saunders, J. Vickers, Y. Zhang, N. De Stefano, J.M. Brady, and P.M. Matthews.
<BR> Advances in functional and structural MR image analysis and
implementation as FSL.
<BR> NeuroImage, 23(S1):208-219, 2004.
<P>[Smith 2002b] S.M. Smith.
<BR> Fast robust automated brain extraction.
<BR> Human Brain Mapping, 17(3):143-155, November 2002.
<P>[Jenkinson 2001] M. Jenkinson and S.M. Smith.
<BR> A global optimisation method for robust affine registration of brain images.
<BR> Medical Image Analysis, 5(2):143-156, June 2001.
<P>[Jenkinson 2002] M. Jenkinson, P.R. Bannister, J.M. Brady, and S.M. Smith.
<BR> Improved optimisation for the robust and accurate linear registration and motion correction of brain images.
<BR> NeuroImage, 17(2):825-841, 2002.
<P>[Zhang 2001] Y. Zhang, M. Brady, and S. Smith.
<BR> Segmentation of brain MR images through a hidden Markov random field model and the expectation maximization algorithm.<BR> IEEE Trans. on Medical Imaging, 20(1):45-57, 2001.
<P>[Bartsch 2004] A.J. Bartsch, N. Bendszus, N. De Stefano, G. Homola, and S. Smith.
<BR> Extending SIENA for a multi-subject statistical analysis of sample-specific cerebral edge shifts: Substantiation of early brain regeneration through abstinence from alcoholism.
<BR> In Tenth Int. Conf. on Functional Mapping of the Human Brain, 2004.
</em></font>
<!-- }}} -->
<!-- }}} -->
<!-- {{{ FSL Tools used -->
<a name="tools"></a><p><hr><H2>FSL Tools used</H2>
This section lists the generic FSL programs that SIENA uses.
<p><a href="../bet/index.html">bet</a> - Brain Extraction Tool. This
automatically removes all non-brain tissue from the image. It can
optionally output the binary brain mask that was derived during this
process, and output an estimate of the external surface of the skull,
for use as a scaling constraint in later registration.
<p><b>pairreg</b>, a script supplied with <a
href="../flirt/index.html">FLIRT</a> - FMRIB's Linear Image
Registration Tool. This script calls FLIRT with a special optimisation
schedule, to register two brain images whilst at the same time using
two skull images to hold the scaling constant (in case the brain has
shrunk over time, or the scanner calibration has changed). The script
first calls FLIRT to register the brains as fully as possible. This
registration is then applied to the skull images, but only the scaling
and skew are allowed to change. This is then applied to the brain
images, and a final pass optimally rotates and translates the brains
to get the best final registration.
<p><a href="../fast/index.html">fast</a> - FMRIB's Automated
Segmentation Tool. This program automatically segments a brain-only
image into different tissue types (normally background, grey matter,
white matter, CSF and other). It also corrects for bias field. It is
used in various ways in the SIENA scripts. Note that both <b>siena</b>
and <b>sienax</b> allow you to choose between segmentation of grey
matter and white matter as separate classes or a single class. It is
important to choose the right option here, depending on whether there
is or is not reasonable grey-white contrast in the image.
<!-- }}} -->
<!-- {{{ SIENA - Two-Time-Point Estimation -->
<a name="siena"></a><p><hr><H2>SIENA - Two-Time-Point Estimation</H2>
<h3>Usage</h3>
<p>The script <a href="siena">siena</a> (see <a
href="siena_usage">usage</a>) is run simply by typing
<p><b>siena <input1_fileroot> <input2_fileroot></b>
<p>where the two input fileroots are images without any filename
extensions. Note that the input fileroot must not contain directory
names - i.e. all must be done within a single directory.
<p>Other options are:
<p><b>-d</b> : debug (don't delete intermediate files)
<p><b>-B "bet options"</b> : if you want to change the BET defaults,
put BET options inside double-quotes after using the -B flag. For
example, to increase the size of brain estimation, use: <code>-B "-f
0.3"</code>
<p><b>-2</b> : two-class segmentation (don't segment grey and white
matter separately) - use this if there is poor grey/white contrast
<p><b>-t2</b>: tell FAST that the input images are T2-weighted and not T1
<p><b>-m</b> : use standard-space masking as well as BET (e.g. if it
is proving hard to get reliable brain segmentation from BET, for
example if eyes are hard to segment out) - register to standard space
in order to use a pre-defined standard-space brain mask
<p><b>-t <t></b>: ignore from t (mm) upwards in MNI152/Talairach space
- if you need to ignore the top part of the head (e.g. if some
subjects have the top missing and you need consistency across
subjects)
<p><b>-b <b></b>: ignore from b (mm) downwards in MNI152/Talairach space; b should probably be -ve
<p><b>-S "siena_diff options"</b> : if you want to send options to the
siena_diff program (that estimates change between two aligned
images), put these options in double-quotes after the -S flag. For
example, to tell siena_diff to run FAST segmentation with an
increased number of iterations, use <code>-S "-s -i 20"</code>
<h3>What the script does</h3>
<p><b>siena</b> carries out the following steps:
<p>Run <b>bet</b> on the two input images, producing as output, for
each input: extracted brain, binary brain mask and skull image. If you
need to call BET with a different threshold than the default of 0.5,
use <b>-f <threshold></b>.
<p>Run <b>siena_flirt</b>, a separate script, to register the two
brain images. This first calls the FLIRT-based registration script
<b>pairreg</b> (which uses the brain and skull images to carry out
constrained registration). It then deconstructs the final transform
into two half-way transforms which take the two brain images into a
space halfway between the two, so that they both suffer the same
amount of interpolation-related blurring. Finally the script produces
a multi-slice gif picture showing the registration quality, with one
transformed image as the background and edges from the other
transformed image superimposed in red.
<p>The final step is to carry out change analysis on the registered
brain images. This is done using the program <b>siena_diff</b>. (In
order to improve slightly the accuracy of the <b>siena_diff</b>
program, a self-calibration script <b>siena_cal</b>, described later,
is run before this.) <b>siena_diff</b> carries out the following
steps:
<UL>
<LI>Transforms original whole head images and brain masks for each
time point into the space halfway between them, using the two halfway
transforms previously generated.
<LI>Combines the two aligned masks using logical OR (if either is 1
then the output is 1).
<LI>The combined mask is used to mask the two aligned head
images, resulting in aligned brain images.
<LI>The change between the two aligned brain images is now estimated,
using the following method (note that options given to the
<b>siena</b> script are passed on to <b>siena_diff</b>): Apply tissue
segmentation to the first brain image. At all points which are
reported as boundaries between brain and non-brain (including internal
brain-CSF boundaries), compute the distance that the brain surface hasmoved between the two time points. This motion of the brain edge
(perpendicular to the local edge) is calulated on the basis of
sub-voxel correlation (matching) of two 1D vectors; these are taken
from the 3D images, a fixed distance either side of the surface point,
and perpendicular to it, and are differentiated before correlation,
allowing some variation in the two original images. Compute mean
perpendicular surface motion and convert to PBVC.
<LI>To make this conversion between mean perpendicular edge motion and
PBVC, it is necessary to assume a certain relationship between real
brain surface area, number of estimated edge points and real brain
volume. This number can be estimated for general images, but will vary
according to slice thickness, image sequence type, etc, causing small
scaling errors in the final PBVC. In order to correct for this,
self-calibration is applied, in which <b>siena</b> calls
<b>siena_cal</b>. This script runs <b>siena_diff</b> on one of the
input images relative to a scaled version of itself, with the scaling
pre-determined (and therefore known). Thus the final PBVC is known in
advance and the estimated value can be compared with this to get a
correction factor for the current image. This is done for both input
images and the average taken, to give a correction factor to be fed
into <b>siena_diff</b>.
</UL>
Note that all output is in the same directory as the input, so this
must be writable by the user. The output files are (assuming the input
images are called "A" and "B"):
<UL>
<LI><code>A_to_B.siena</code> the output information from the <b>siena</b> script.
<LI><code>A_halfwayto_B_render</code> a colour-rendered image of edge motion
superimposed on the halfway A image.
<LI><code>B_regto_A.gif</code> a gif image showing the results of the registration,
using one transformed image as the background and the other as the
coloured edges foreground.
<LI><code>A_to_B.mat</code> the transformation taking A to B, using the brain and
skull images.
<LI><code>B_to_A.mat</code> the transformation taking B to A, using the brain and
skull images.
<LI><code>A_halfwayto_B.mat</code> and <code>B_halfwayto_A.mat</code> the transformations taking
the images to the halfway positions.
</UL>
<!-- }}} -->
<!-- {{{ SIENAX - Single-Time-Point Estimation -->
<a name="sienax"></a><p><hr><H2>SIENAX - Single-Time-Point Estimation</H2>
<h3>Usage</h3>
<p>The script <a href="sienax">sienax</a> (see <a
href="sienax_usage">usage</a>) is run simply by typing
<p><b>sienax <input_fileroot></b>
<p>where the input fileroot is an image without any filename
extension. Note that the input fileroot must not contain directory
names - i.e. all must be done within a single directory.
<p>Other options are:
<p><b>-d</b> : debug (don't delete intermediate files)
<p><b>-B "bet options"</b> : if you want to change the BET defaults,
put BET options inside double-quotes after using the -B flag. For
example, to increase the size of brain estimation, use: <code>-B "-f
0.3"</code>
<p><b>-2</b> : two-class segmentation (don't segment grey and white
matter separately) - use this if there is poor grey/white contrast
<p><b>-t2</b>: tell FAST that the input images are T2-weighted and not T1
<p><b>-t <t></b>: ignore from t (mm) upwards in MNI152/Talairach space
- if you need to ignore the top part of the head (e.g. if some
subjects have the top missing and you need consistency across
subjects)
<p><b>-b <b></b>: ignore from b (mm) downwards in MNI152/Talairach space; b should probably be -ve
<p><b>-r</b>: tell SIENAX to estimate "regional" volumes as well as
global; this produces peripheral cortex GM volume (3-class
segmentation only) and ventricular CSF volume
<p><b>-lm <mask></b>: use a lesion (or lesion+CSF) mask to
remove incorrectly labelled "grey matter" voxels
<p><b>-S "FAST options"</b> : if you want to change the segmentation
defaults, put FAST options inside double-quotes after using the -S
flag. For example, to increase the number of segmentation
iterations use: <code>-S "-i 20"</code>
<h3>What the script does</h3>
<p><b>sienax</b> carries out the following steps:
<UL>
<LI>Run <b>bet</b> on the single input image, outputting the
extracted brain, and the skull image. If you need to call BET with a
different threshold than the default of 0.5, use <b>-f
<threshold></b>.
<LI>Run <b>pairreg</b> (which uses the brain and skull images to
carry out constrained registration); the MNI152 standard brain is
the target (reference), using brain and skull images derived from
the MNI152. Thus, as with two-time-point atrophy, the brain is
registered (this time to the standard brain), again using the skull
as the scaling constraint. Thus brain tissue volume (estimated
below) will be relative to a "normalised" skull size. (Ignore the
"WARNING: had difficulty finding robust limits in histogram"
message; this appears because FLIRT isn't too happy with the unusual
histograms of skull images, but is nothing to worry about in this
context.) Note that all later steps are in fact carried out on the
original (but stripped) input image, not the registered input image;
this is so that the original image does not need to be resampled
(which introduces blurring). Instead, to make use of the
normalisation described above, the brain volume (estimated by the
segmentation step described below) is scaled by a scaling factor
derived from the normalising transform, before being reported as the
final normalised brain volume.
<LI>A standard brain image mask, (derived from the MNI152 and
slightly dilated) is transformed into the original image space (by
inverting the normalising transform found above) and applied to the
brain image. This helps ensure that the original brain extraction
does not include artefacts such as eyeballs.
<LI>Segmentation is now run on the masked brain using <b>fast</b>. If there is reasonable grey-white contrast, grey
matter and white matter volumes are reported separately, as well as
total brain volume (this is the default behaviour). Otherwise
(i.e. if <b>sienax</b> was called with the <b>-2</b> option), just
brain/CSF/background segmentation is carried out, and only brain
volume is reported. Before reporting, all volumes are scaled by the
normalising scaling factor, as described above, so that all
subjects' volumes are reported relative to a normalised skull size.
</UL>
Note that all output is in the same directory as the input, so this
must be writable by the user. The output files are (assuming the input
image is called "A"):
<UL>
<LI><code>A.sienax</code> the output information from the <b>sienax</b> script.
<LI><code>A_render</code> a colour-rendered image showing the segmentation
output superimposed on top of the original image.
</UL>
<!-- }}} -->
<!-- {{{ Voxelwise SIENA Statistics -->
<a name="sienar"></a><p><hr><H2>Voxelwise SIENA Statistics</H2>
<p>We have extended SIENA to allow the voxelwise statistical analysis
of atrophy across subjects. This takes a SIENA-derived edge "flow
image" (edge displacement between the timepoints) for each subject,
warps these to align with a standard-space edge image and then carries
out voxelwise cross-subject statistical analysis to identify brain
edge points which, for example, are signficantly atrophic for the
group of subjects as a whole, or where atrophy correlates
significantly with age or disease progression.
<p>In order to carry out voxelwise SIENA statistics, do the following:
<UL>
<LI> Run <br><code>siena A B</code><br> on all subjects'
two-timepoints data (here <code>A</code> and <code>B</code>).
<LI> For each subject run <br><code>siena_flow2std A B</code><br> -
this runs flirt to generate the transform to standard space (if it
doesn't already exist), takes the edge flow (atrophy) image generated
by <code>siena</code>, dilates this several times (to "thicken" this edge
flow image), transforms to standard space, and masks with a standard
space edge mask. It then smooths this with a Gaussian filter of
half-width 10mm before remasking; you may want to edit this or even
remove this step - find the call to <code>ip</code> in
$FSLDIR/bin/siena_flow2std and change the <code>10</code> or just remove
that line.
<LI> All subjects will now have an edge flow image in standard edge
space called <code>A_to_B_flow_to_std</code>. Merge these into a single 4D
image; for example, if each subject's analysis has so far been carried
out in a subdirectory called siena_subject_*, where the * could be
subject ID or name, use a command such as:<br><code>
avwmerge -t flow_all_subjects `imglob -oneperimage siena_subject_*/A_to_B_flow_to_std*`</code><br>
Note: it is very important that the order that the subjects appear in
this command matches the order you intend when you then create the
design matrix!
<LI> You are now ready to carry out the cross-subject statistics. We
recommend using <a href="../randomise/index.html">randomise</a> for
this, as the above steps are very unlikely to generate nice Gaussiandistributions in the data. You will need to generate a FEAT-style
design matrix <code>design.mat</code> and contrasts file
<code>design.con</con>.
</UL>
<!-- }}} -->
<!-- {{{ Manual Segmentation Correction -->
<!-- <hr><H2>Manual Segmentation Correction</H2> -->
<!-- <p>If you want to manually edit the segmentation image, to correct for -->
<!-- any segmentation errors, you should follow the steps below. The -->
<!-- following assumes the use of MEDx to edit images. The main steps -->
<!-- described are for correcting <b>sienax</b>, followed by a description -->
<!-- of how to adapt this procedure for <b>siena</b>. -->
<!-- <UL> -->
<!-- <LI>Load the ...seg.hdr image into MEDx. -->
<!-- <LI>In lightbox mode, make sure the dimensionality is 3D. -->
<!-- <LI>In "look-at-voxel-values" mode rather than the "graphics-edit" -->
<!-- mode (controlled by the button to the right of the 3D/2D button), -->
<!-- check the intensities of the classes of interest - for example, -->
<!-- 0=background, 1=CSF, 2=grey, 3=white. -->
<!-- <LI>Using <b>Toolbox->Measurement->Statistics</b>, set the lower and -->
<!-- upper thresholds to give the current count of G+W (eg lower=1.5, -->
<!-- upper=3.5) and measure the <b>Area</b> (in fact the volume). This is -->
<!-- slightly different (and less accurate) than that reported by fast -->
<!-- as it does not directly take into account partial volume effect. -->
<!-- <LI>In lightbox mode, change dimensionality from 3D to 2D. -->
<!-- <LI>Using the pencil tool from the toolbox, outline all regions of -->
<!-- one particular type that you wish to change to a different type. -->
<!-- <LI>Use <b>Folder->Graphic->Select All->2D Graphics</b> to highlight -->
<!-- all graphics that you've just drawn. Use -->
<!-- <b>Folder->Graphic->Grouping->Derive 3D Contour</b> to convert these -->
<!-- into one 3D object. -->
<!-- <LI>Bring up the image calculator and enter the intensity of the -->
<!-- class that you wish to set all highlighted voxels to (eg 2). Press -->
<!-- <b>OK</b>. Delete the 3D contour with -->
<!-- <b>Folder->Graphic->Delete</b>, and reset the display range using -->
<!-- <b>Folder->Display->Display Range->Compute</b>. -->
<!-- <LI>Now repeat the last 4 steps if you wish to also change other -->
<!-- areas to a different class type to that which you have just set. -->
<!-- <LI>Now re-run the measurement of "<b>Area</b>" using -->
<!-- statistics. Subtract the two volumes, multiply by the VSCALING value -->
<!-- (from the .sienax output file) and amend the BRAIN volume reported -->
<!-- at the end of this file. -->
<!-- </UL> -->
<!-- If you want to amend the segmentation carried out within <b>siena</b>, -->
<!-- carry out the above (without bothering with any of the statistics -->
<!-- steps) and save the image, overwriting the original segmentation -->
<!-- output. Now if you re-run <b>siena</b> the edited segmentation result -->
<!-- will be used (as <b>siena</b> doesn't re-run segmentation if a -->
<!-- segmentation output image is already in existence). -->
<!-- }}} -->
<!-- {{{ end -->
<p><HR><FONT SIZE=1>Copyright © 2000-2004, University of
Oxford. Written by <A
HREF="http://www.fmrib.ox.ac.uk/~steve/index.html">S. Smith</A>.</FONT>
<br><br><br><br><br><br><br><br><br><br><br><br><br><br><br><br><br>
<br><br><br><br><br><br><br><br><br><br><br><br><br><br><br><br><br>
<br><br><br><br><br><br><br><br><br><br><br><br><br><br><br><br><br>
</BODY></HTML>
<!-- }}} -->