|
Tutorial: detecting crossing fibresTutorials.PigBrainTutorial HistoryHide minor edits - Show changes to markup August 06, 2010, at 09:26 PM
by
- Changed line 11 from:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the available formats. The to:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the available formats. The August 06, 2010, at 09:24 PM
by
- Changed line 34 from:
[@pdview -inputfile pig_dt_eig_sys.Bdouble -scalarfile pig_fa.hdr]
to:
pdview -inputfile pig_dt_eig_sys.Bdouble -scalarfile pig_fa.hdr
August 06, 2010, at 09:24 PM
by
- Changed lines 11-12 from:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the available formats. The to:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the available formats. The Changed lines 32-36 from:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/fa05.png to:
This command produces several Analyze images, and some Camino raw data files. All output is prepended with pig_. See the man page for analyzedti for more details. We can check the reconstruction using [@pdview -inputfile pig_dt_eig_sys.Bdouble -scalarfile pig_fa.hdr]
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/pig_pdview.png Changed line 42 from:
to:
Changed line 46 from:
to:
August 06, 2010, at 09:11 PM
by
- Changed lines 28-29 from:
ls *.hdr > imagelist.txt
to:
ls NEX*.hdr > imagelist.txt
Changed lines 35-70 from:
to:
Diffusion tensor plotThe command sfplot creates plots of diffusion MRI reconstructions over slices. In particular, it can create plots of the diffusion tensor over a slice. It can also create plots of other reconstructed features, such as the PAS or q-ball ODF; see for example the Parallel PASMRI case study or the sfplot man page. The following commands create a plot of the DT over slice 5 of the pig brain data set. We start by creating a fractional anisotropy map to use as background for the plot: \\ August 06, 2010, at 09:03 PM
by - August 06, 2010, at 09:03 PM
by - August 06, 2010, at 09:03 PM
by
- Changed lines 7-8 from:
The data set in this case study comes from a post-mortem porcine brain and was kindly provided by Tim Dyrby from The Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital, Hvidovre, Denmark. The data set is intended for demonstration and testing of the Camino software package only. If you wish to use this data for other purposes, such as publications, other web sites or commercial enterprises, please contact Tim at (timd|at|drcmr.dk) before doing so. You can download the data set . The acquisition uses a spherical acquisition scheme with 61 unique gradient directions. The gradient directions come from the file camino/PointSets/Elec061.txt and minimize the electrostatic energy for 61 pairs of equal-and-opposite point electrical charges free to move on the surface of the sphere. The procedure outlined in [Jansons and Alexander, Inverse Problems, Vol 19, pp 1031-1046, 2003] computed them. The diffusion-weighting gradient strength G = 61 mT m-1; pulse onset separation DELTA = 30ms; pulse width delta = 23ms; TE = 60ms; b = 3146 s mm-2; NEX=2. The protocol also includes 3 acquisitions with no diffusion weighting. The image has 10 slices with in-plane resolution 128x128; voxel size 0.5x0.5x0.5 mm3. You can download the data here. to:
The data set in this case study comes from a post-mortem porcine brain and was kindly provided by Tim Dyrby from The Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital, Hvidovre, Denmark. The data set is intended for demonstration and testing of the Camino software package only. If you wish to use this data for other purposes, such as publications, other web sites or commercial enterprises, please contact Tim at (timd|at|drcmr.dk) before doing so. You can download the data set . The acquisition uses a spherical acquisition scheme with 61 unique gradient directions. The gradient directions come from the file camino/PointSets/Elec061.txt and minimize the electrostatic energy for 61 pairs of equal-and-opposite point electrical charges free to move on the surface of the sphere. The procedure outlined in [Jansons and Alexander, Inverse Problems, Vol 19, pp 1031-1046, 2003] computed them. The diffusion-weighting gradient strength G = 61 mT m-1; pulse onset separation DELTA = 30ms; pulse width delta = 23ms; TE = 60ms; b = 3146 s mm-2; NEX=2. The protocol also includes 3 acquisitions with no diffusion weighting. The image has 10 slices with in-plane resolution 128x128; voxel size 0.5x0.5x0.5 mm3. You can download the data here. Changed lines 11-24 from:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the format. The simple perl script
The first argument is the point set file; the second is the wavenumber q = gamma*delta*G = 2.6751987E8 * 0.023 * 0.061, where gamma is the gyromagnetic ratio of protons in water; the third argument is the number of b=0 images and the last is the diffusion time. Here we use the standard approximation of the diffusion time for finite pulse widths t = DELTA - delta/3 = 0.03 - 0.23/3. This is the resulting . A scheme file specifies the gradient directions along with the zero directions. The zero directions should always be positioned before the non-zero directions. Data FileEach acquisition from the scanner is in an analyze format file to:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the available formats. The pointset2scheme -inputfile pigPoints.txt -version STEJSKALTANNER -dwparams 0.061 0.03 0.023 0.06 > PigBrain_ST.scheme
The "version" argument tells the program that we want to describe a Stejskal-Tanner sequence, which uses two rectangular gradient pulses of equal duration. The "dwparams" argument is followed by the parameters of the sequence: gradient strength (0.061 T / m), pulse separation (0.03 s), pulse duration (0.023 s), and echo time (0.06 s). Alternatively, we could compute the b-value for the scheme, 3146 s/mm^2, and created a less detailed scheme file: pointset2scheme -inputfile pigPoints.txt -version BVECTOR -dwparams 3146 > PigBrain_BVEC.scheme
For most applications, different scheme files are equivalent as long as they describe the same b-value. Some programs, like the monte-carlo diffusion simulation, require a more detailed description of the pulse sequence. Changed lines 24-68 from:
To fit the diffusion tensor in each voxel we can use dtfit or modelfit:
or equivalently:
Simple tensor statisticsFrom the diffusion tensor, we can compute simple statistics such as Trace(D), fractional anisotropy and principal directions. The commands trd and fa compute the first two:
A simple way to look at the results is to load the data files into matlab and display with imshow (use image instead if you do not have the image-processing toolbox) or write out slice by slice:
to:
Each acquisition from the scanner is in an analyze format file To fit the diffusion tensor in each voxel we can use analyzedti: ls *.hdr > imagelist.txt
analyzedti imagelist.txt ./pig_ PigBrain_ST.scheme
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/fa05.png \\ December 09, 2009, at 03:11 PM
by
- added title, link to pig brain dataChanged lines 1-2 from:
Tutorial: detecting crossing fibresto:
(:title Tutorial: detecting crossing fibres:) Changed line 7 from:
The data set in this case study comes from a post-mortem porcine brain and was kindly provided by Tim Dyrby from The Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital, Hvidovre, Denmark. The data set is intended for demonstration and testing of the Camino software package only. If you wish to use this data for other purposes, such as publications, other web sites or commercial enterprises, please contact Tim at (timd|at|drcmr.dk) before doing so. You can download the data set . The acquisition uses a spherical acquisition scheme with 61 unique gradient directions. The gradient directions come from the file camino/PointSets/Elec061.txt and minimize the electrostatic energy for 61 pairs of equal-and-opposite point electrical charges free to move on the surface of the sphere. The procedure outlined in [Jansons and Alexander, Inverse Problems, Vol 19, pp 1031-1046, 2003] computed them. The diffusion-weighting gradient strength G = 61 mT m-1; pulse onset separation DELTA = 30ms; pulse width delta = 23ms; TE = 60ms; b = 3146 s mm-2; NEX=2. The protocol also includes 3 acquisitions with no diffusion weighting. The image has 10 slices with in-plane resolution 128x128; voxel size 0.5x0.5x0.5 mm3. to:
The data set in this case study comes from a post-mortem porcine brain and was kindly provided by Tim Dyrby from The Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital, Hvidovre, Denmark. The data set is intended for demonstration and testing of the Camino software package only. If you wish to use this data for other purposes, such as publications, other web sites or commercial enterprises, please contact Tim at (timd|at|drcmr.dk) before doing so. You can download the data set . The acquisition uses a spherical acquisition scheme with 61 unique gradient directions. The gradient directions come from the file camino/PointSets/Elec061.txt and minimize the electrostatic energy for 61 pairs of equal-and-opposite point electrical charges free to move on the surface of the sphere. The procedure outlined in [Jansons and Alexander, Inverse Problems, Vol 19, pp 1031-1046, 2003] computed them. The diffusion-weighting gradient strength G = 61 mT m-1; pulse onset separation DELTA = 30ms; pulse width delta = 23ms; TE = 60ms; b = 3146 s mm-2; NEX=2. The protocol also includes 3 acquisitions with no diffusion weighting. The image has 10 slices with in-plane resolution 128x128; voxel size 0.5x0.5x0.5 mm3. You can download the data here. October 22, 2009, at 02:33 AM
by
- Changed lines 151-167 from:
to:
October 22, 2009, at 02:30 AM
by
- Changed line 127 from:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/PigBrain05_PAS.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/PigBrain05_PAS.png \\ October 22, 2009, at 02:28 AM
by
- Changed line 142 from:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/VC_ThreshSelect.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/VC_ThreshSelect.png\\ October 22, 2009, at 02:13 AM
by
- Changed line 19 from:
Each acquisition from the scanner is in an analyze format file NEX2_s_20060127_001_semsdw_15_image??, where ?? goes from 01 to 64; 01-03 are the b=0 images and 04-64 are the diffusion-weighted images with gradients in the same order as the scheme file. Each file stores image intensities as little-endian 2-byte short integers. We must generate a data file in camino format, which requires switching to big endian representation and combining all the analyze files into a single file with voxel ordering (see camino/man/man1/camino.1). Here is the command: \\ to:
Each acquisition from the scanner is in an analyze format file Changed lines 85-105 from:
to:
Changed line 127 from:
files/pig/PigBrain05_PAS.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/PigBrain05_PAS.png \\ Changed line 142 from:
files/pig/VC_ThreshSelect.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/VC_ThreshSelect.png \\ Changed line 147 from:
to:
Changed line 149 from:
The output of this command has type int and every voxel contains either -1 for background or 0, 2 or 4 indicating the classification order. A nice visualization of the results highlights the fibre crossings on the fractional anisotropy map. This is simple in matlab: \\ to:
The output of this command has type int and every voxel contains either -1 for background or 0, 2 or 4 indicating the classification order. A nice visualization of the results highlights the fibre crossings on the fractional anisotropy map. This is simple in Matlab: \\ Changed line 152 from:
files/pig/favc05.png to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/favc05.png October 22, 2009, at 02:08 AM
by
- Changed line 58 from:
>> imwrite(flipud(trd(:,:,i)'), sprintf('trd%02i.png', i)); \\
to:
Changed line 68 from:
>> imwrite(flipud(fa(:,:,i)'), sprintf('fa%02i.png', i)); \\
to:
Changed line 72 from:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/trd05.pngfiles/pig/fa05.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/fa05.png \\ October 22, 2009, at 02:07 AM
by
- Changed lines 54-55 from:
to:
Changed line 58 from:
>> imwrite(flipud(trd(:,:,i)'), sprintf('trd%02i.png', i)); \\
to:
>> imwrite(flipud(trd(:,:,i)'), sprintf('trd%02i.png', i)); \\
October 22, 2009, at 02:06 AM
by
- Changed lines 50-53 from:
to:
Changed line 57 from:
to:
>> imwrite(flipud(trd(:,:,i)'), sprintf('trd%02i.png', i)); \\
Changed line 61 from:
to:
Changed line 67 from:
to:
>> imwrite(flipud(fa(:,:,i)'), sprintf('fa%02i.png', i)); \\
October 22, 2009, at 02:05 AM
by
- Changed lines 50-63 from:
to:
Changed lines 65-68 from:
>> imwrite(flipud(fa(:,:,i)'), sprintf('fa%02i.png', i)); >> end \\
to:
October 22, 2009, at 02:02 AM
by
- Changed line 44 from:
to:
October 22, 2009, at 02:01 AM
by
- Changed lines 44-45 from:
@@trd < PigBrainDT.Bdouble > PigBrainTrD.Bdouble fa < PigBrainDT.Bdouble > PigBrainFA.Bdouble@@ to:
October 22, 2009, at 02:01 AM
by
- Changed lines 44-45 from:
to:
@@trd < PigBrainDT.Bdouble > PigBrainTrD.Bdouble fa < PigBrainDT.Bdouble > PigBrainFA.Bdouble@@ Changed line 49 from:
to:
October 22, 2009, at 01:59 AM
by
- Changed line 48 from:
to:
October 22, 2009, at 01:58 AM
by
- Changed lines 11-13 from:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the format. The simple perl script camino/PointSets/PointSetToScheme converts pigPoints.txt to a scheme file: to:
Before we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the format. The simple perl script
Changed lines 27-33 from:
To fit the diffusion tensor in each voxel we can use dtfit or modelfit: \\ to:
To fit the diffusion tensor in each voxel we can use dtfit or modelfit:
or equivalently:
Changed lines 35-36 from:
to:
Both commands give the same output. Note that the default output data type is big-endian doubles. The modelfit program has more options (see the modelfit man page) than dtfit, such as thresholding the b=0 measurements to remove the noisy background you can see in many of the images below, like this: \\ Changed line 37 from:
to:
Changed lines 39-43 from:
Both commands give the same output. Note that the default output data type is big-endian doubles. The modelfit program has more options (see the modelfit man page) than dtfit, such as thresholding the b=0 measurements to remove the noisy background you can see in many of the images below, like this: to:
Changed lines 42-43 from:
From the diffusion tensor, we can compute simple statistics such as Trace(D), fractional anisotropy and principal directions. The commands trd and fa compute the first two: to:
From the diffusion tensor, we can compute simple statistics such as Trace(D), fractional anisotropy and principal directions. The commands trd and fa compute the first two:
A simple way to look at the results is to load the data files into matlab and display with imshow (use image instead if you do not have the image-processing toolbox) or write out slice by slice:
>> imwrite(flipud(fa(:,:,i)'), sprintf('fa%02i.png', i)); >> end \\
Deleted lines 68-70:
A simple way to look at the results is to load the data files into matlab and display with imshow (use image instead if you do not have the image-processing toolbox) or write out slice by slice: Changed line 70 from:
files/pig/trd05.pngfiles/pig/fa05.png \\ to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/trd05.pngfiles/pig/fa05.png \\ Changed lines 74-75 from:
to:
Changed line 81 from:
We can generate colour-coded direction maps easily from the output in matlab: \\ to:
We can generate colour-coded direction maps easily from the output in Matlab: \\ Changed line 86 from:
files/pig/pds05.png to:
http://www.cs.ucl.ac.uk/research/medic/camino/tutorials/files/pig/pds05.png September 04, 2009, at 04:32 PM
by
- Added lines 1-106:
Tutorial: detecting crossing fibresThis tutorial uses Camino to generate fractional-anisotropy, colour-coded direction and voxel-classification maps. The voxel classification maps indicate which voxels exhibit isotropic diffusion, anisotropic Gaussian diffusion (adequately modelled by a single diffusion tensor) and non-Gaussian displacements (the diffusion tensor model fails), which are likely fibre crossings. DataThe data set in this case study comes from a post-mortem porcine brain and was kindly provided by Tim Dyrby from The Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital, Hvidovre, Denmark. The data set is intended for demonstration and testing of the Camino software package only. If you wish to use this data for other purposes, such as publications, other web sites or commercial enterprises, please contact Tim at (timd|at|drcmr.dk) before doing so. You can download the data set . The acquisition uses a spherical acquisition scheme with 61 unique gradient directions. The gradient directions come from the file camino/PointSets/Elec061.txt and minimize the electrostatic energy for 61 pairs of equal-and-opposite point electrical charges free to move on the surface of the sphere. The procedure outlined in [Jansons and Alexander, Inverse Problems, Vol 19, pp 1031-1046, 2003] computed them. The diffusion-weighting gradient strength G = 61 mT m-1; pulse onset separation DELTA = 30ms; pulse width delta = 23ms; TE = 60ms; b = 3146 s mm-2; NEX=2. The protocol also includes 3 acquisitions with no diffusion weighting. The image has 10 slices with in-plane resolution 128x128; voxel size 0.5x0.5x0.5 mm3. Scheme FileBefore we can process the data, we must create a scheme file that tells Camino about the acquisition sequence. The scheme file is a text file; the Camino man page (camino/man/man1/camino.1) specifies the format. The simple perl script camino/PointSets/PointSetToScheme converts pigPoints.txt to a scheme file: Data FileEach acquisition from the scanner is in an analyze format file NEX2_s_20060127_001_semsdw_15_image??, where ?? goes from 01 to 64; 01-03 are the b=0 images and 04-64 are the diffusion-weighted images with gradients in the same order as the scheme file. Each file stores image intensities as little-endian 2-byte short integers. We must generate a data file in camino format, which requires switching to big endian representation and combining all the analyze files into a single file with voxel ordering (see camino/man/man1/camino.1). Here is the command: Fit the diffusion tensorTo fit the diffusion tensor in each voxel we can use dtfit or modelfit: Simple tensor statisticsFrom the diffusion tensor, we can compute simple statistics such as Trace(D), fractional anisotropy and principal directions. The commands trd and fa compute the first two: Diffusion tensor plotThe command sfplot creates plots of diffusion MRI reconstructions over slices. In particular, it can create plots of the diffusion tensor over a slice. It can also create plots of other reconstructed features, such as the PAS or q-ball ODF; see for example the Parallel PASMRI case study or the sfplot man page. The following commands create a plot of the DT over slice 5 of the pig brain data set. We start by creating a fractional anisotropy map to use as background for the plot: Voxel ClassificationThe command |