|
QBallTutorialTutorials.QBallTutorial HistoryHide minor edits - Show changes to markup June 24, 2010, at 02:54 PM
by
- Changed line 75 from:
to:
February 25, 2010, at 10:50 AM
by
- Changed lines 71-72 from:
We choose the value for -bgthresh by thresholding the b0 image such that the background is completely masked. This can be easily be done in Matlab. to:
We choose the value for -bgthresh by thresholding the b0 image such that the background is masked while brain voxels remain. Use any image viewer (MRIcro/ITKsnap/Matlab) to compare foreground and background intensities in the b0 image and pick a value that lies between the two. Changed lines 98-101 from:
Camino also implements the spherical harmonic (SH) Q-Ball reconstructions of Descoteaux [2] and Mukherjee [3]. We will now create an SH representation of the Q-Ball ODFs and use
to:
Camino also implements the spherical harmonic (SH) Q-Ball reconstructions of Descoteaux [3] and Mukherjee [4]. We will now create an SH representation of the Q-Ball ODFs and use
Changed lines 119-120 from:
[2] M. Descoteaux. A fast and robust ODF estimation algorithm in Q-Ball imaging, Biomedical Imaging: Macro to Nano. 3rd IEEE International Symposium, 81-84, 2006 to:
[2] D. C. Alexander. Multiple fibre reconstruction algorithms for diffusion MRI, Annals of the New York Academy of Sciences 1046:113-133 2005. February 25, 2010, at 10:46 AM
by
- Changed line 49 from:
to:
February 25, 2010, at 10:45 AM
by
- Changed lines 1-6 from:
Tutorial: Q-BallThis tutorial illustrates how to use Tuch's Q-Ball [1] method to reconstruct multiple fibre directions in single voxels. Q-Ball is a linear reconstruction method capable of resolving complex sub-voxel structure and, in particlular, capable of resolving multiple fibre directions in a single voxel. In contrast to the PAS method used in the pig brain tutorial, Q-Ball is a linear reconstruction method and as such is computationally less demanding. This tutorial demonstrates Camino command pipelines that generate Q-Ball images of reconstructed fibre orientation distribution functions (ODF) and generate statistics from these reconstructions. to:
Tutorial: Q-Ball ImagingThis tutorial illustrates how to use Q-Ball imaging [1] to reconstruct multiple fibre directions in single voxels. Q-Ball is a linear reconstruction method capable of resolving complex sub-voxel structure and, in particlular, capable of resolving multiple fibre directions in a single voxel. In contrast to the PAS method used in the pig brain tutorial, Q-Ball is a linear reconstruction method and as such is computationally less demanding, although it is generally less sensitive to crossing fibres. This tutorial demonstrates Camino command pipelines that generate images of Q_Ball orientation distribution functions (ODF) and generate statistics from them. Changed line 9 from:
The data used in this study is the same as that used in the pig brain tutorial. The pig brain tutorial discusses the neccessary steps to preprocess data into voxel order, the format expected by Camino. If your data is in Analyze format, the program to:
The data used in this study is the same as that used in the pig brain tutorial. The pig brain tutorial discusses the necessary steps to preprocess data into voxel order, as expected by Camino. If your data is in Analyze format, the program Changed lines 19-22 from:
This section shows how to use Camino to create a spherical radial basis function (sRBF) representation of Q-Ball ODFs in each voxel of a brain volume. We will then visualize the ODFs of one slice of brain data using Prior to calculating the Q-Ball ODFs, we will perform a diffusion tensor reconstruction of the data. There are two reasons for doing this. Firstly, we use the DT fit to test that none of the directions are flipped using the commands: to:
This section shows how to use Camino to create a spherical radial basis function (sRBF) representation of Q-Ball ODFs in each voxel of a brain volume. (This is Tuch's original approach in [1]; later we will look at spherical harmonic Q-Ball, which is more economical.) We will then visualize the ODFs of one slice of brain data using Prior to calculating the Q-Ball ODFs, we will perform a diffusion tensor reconstruction of the data. There are two reasons for doing this. First, we use the DT fit to test that none of the directions are flipped using the commands: Changed lines 37-38 from:
The first stage required for Q-Ball reconstruction is the generation of the Q-Ball reconstruction matrix. This matrix transforms the data from a sphere in q-space to a sphere in real space (the ODF). To calculate the Q-Ball reconstruction matrix we use to:
The first stage required for Q-Ball reconstruction is the generation of the Q-Ball reconstruction matrix. This matrix transforms the data from a sphere in q-space to the ODF. To calculate the Q-Ball reconstruction matrix we use Changed lines 45-47 from:
The extra options can be changed to adapt the behaviour of the algorithm, see [1] for details. If the program has run successfully, you will see information about the basis functions used as well as their settings on the command console. Record this information for future reference as you will need it later. We check that the settings (i.e. the number ODF basis functions and their widths) give reasonable results by generating some test ODFs using synthetic data from
to:
The extra options can be changed to adapt the behaviour of the algorithm, see [1,2] for details. If the program has run successfully, you will see information about the basis functions used as well as their settings on the command console. Record this information for future reference as you will need it later. We can check that the settings (i.e. the number ODF basis functions and their widths) give reasonable results by generating some test ODFs using synthetic data from
October 22, 2009, at 02:42 AM
by
- Changed line 116 from:
[3] C.R. Hess et al. Q-ball reconstruction of multimodal fiber orientations using the spherical harmonic, Journal Magnetic Resonance Imaging, 56:104-117, 2006. \\ to:
[3] C.R. Hess et al. Q-ball reconstruction of multimodal fiber orientations using the spherical harmonic, Journal Magnetic Resonance Imaging, 56:104-117, 2006. October 22, 2009, at 02:41 AM
by
- Changed lines 3-4 from:
This tutorial illustrates how to use Tuch's Q-Ball [1] method to reconstruct multiple fibre directions in single voxels. Q-Ball is a linear reconstruction method capable of resolving complex sub-voxel structure and, in particlular, capable of resolving multiple fibre directions in a single voxel. In contrast to the PAS method used in the pig brain tutorial?, Q-Ball is a linear reconstruction method and as such is computationally less demanding. to:
This tutorial illustrates how to use Tuch's Q-Ball [1] method to reconstruct multiple fibre directions in single voxels. Q-Ball is a linear reconstruction method capable of resolving complex sub-voxel structure and, in particlular, capable of resolving multiple fibre directions in a single voxel. In contrast to the PAS method used in the pig brain tutorial, Q-Ball is a linear reconstruction method and as such is computationally less demanding. Changed line 9 from:
The data used in this study is the same as that used in the pig brain tutorial?. The pig brain tutorial discusses the neccessary steps to preprocess data into voxel order, the format expected by Camino. If your data is in Analyze format, the program to:
The data used in this study is the same as that used in the pig brain tutorial. The pig brain tutorial discusses the neccessary steps to preprocess data into voxel order, the format expected by Camino. If your data is in Analyze format, the program Changed lines 15-16 from:
Scan parameters and gradient directions are contained in the camino schemefile for the dataset. Schemefiles for the pig brain dataset are also available in the pig brain tutorial?. to:
Scan parameters and gradient directions are contained in the camino schemefile for the dataset. Schemefiles for the pig brain dataset are also available in the pig brain tutorial. Changed lines 23-25 from:
files/qball/dtCheck_Right.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/qball/dtCheck_Right.png \\ Changed line 30 from:
files/qball/dtCheck_Wrong.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/qball/dtCheck_Wrong.png \\ Changed line 53 from:
files/qball/testODFs.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/qball/testODFs.png \\ Changed lines 58-59 from:
to:
Changed lines 70-73 from:
To create a colour-coded image using
to:
To create a colour-coded image using
Changed lines 81-83 from:
files/qball/pigSFPLOT.png \\ to:
Changed line 109 from:
files/qball/qball_sh_pdview.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/qball/qball_sh_pdview.png \\ September 04, 2009, at 04:42 PM
by
- Added lines 1-102:
Tutorial: Q-BallThis tutorial illustrates how to use Tuch's Q-Ball [1] method to reconstruct multiple fibre directions in single voxels. Q-Ball is a linear reconstruction method capable of resolving complex sub-voxel structure and, in particlular, capable of resolving multiple fibre directions in a single voxel. In contrast to the PAS method used in the pig brain tutorial?, Q-Ball is a linear reconstruction method and as such is computationally less demanding. This tutorial demonstrates Camino command pipelines that generate Q-Ball images of reconstructed fibre orientation distribution functions (ODF) and generate statistics from these reconstructions. Data & Scheme FileThe data used in this study is the same as that used in the pig brain tutorial?. The pig brain tutorial discusses the neccessary steps to preprocess data into voxel order, the format expected by Camino. If your data is in Analyze format, the program Scan parameters and gradient directions are contained in the camino schemefile for the dataset. Schemefiles for the pig brain dataset are also available in the pig brain tutorial?. Generating a Q-Ball ODF mapThis section shows how to use Camino to create a spherical radial basis function (sRBF) representation of Q-Ball ODFs in each voxel of a brain volume. We will then visualize the ODFs of one slice of brain data using Prior to calculating the Q-Ball ODFs, we will perform a diffusion tensor reconstruction of the data. There are two reasons for doing this. Firstly, we use the DT fit to test that none of the directions are flipped using the commands:
files/qball/dtCheck_Right.png files/qball/dtCheck_Wrong.png If we examine an axial slice in the centre of the brain volume, we should see that the principal directions of the diffusion tensor follow the white-matter tracts (see fig 1a). However, if one of the directions is flipped, for example the y direction, then we will see an image similar to that of fig 1b. We also use the diffusion tensor reconstruction to create the FA map for the background image. This is achieved using the
The first stage required for Q-Ball reconstruction is the generation of the Q-Ball reconstruction matrix. This matrix transforms the data from a sphere in q-space to a sphere in real space (the ODF). To calculate the Q-Ball reconstruction matrix we use
This command uses sensible defaults for the various parameters of the Q-Ball algorithm and is equivalent to
The extra options can be changed to adapt the behaviour of the algorithm, see [1] for details. If the program has run successfully, you will see information about the basis functions used as well as their settings on the command console. Record this information for future reference as you will need it later. We check that the settings (i.e. the number ODF basis functions and their widths) give reasonable results by generating some test ODFs using synthetic data from
To visualize this test data, we can create a basic greyscale image using sfplot:
files/qball/testODFs.png To view the image on a UNIX machine with Image Magick installed, use the command
Once you have found settings that give reasonable results, you calculate the ODFs for each voxel of the pig brain dataset using:
We choose the value for -bgthresh by thresholding the b0 image such that the background is completely masked. This can be easily be done in Matlab. To create a colour-coded image using
To create a colour-coded image, we add the flag
This image can be viewed in
files/qball/pigSFPLOT.png Using
|