questions > dcm2nii bvals from Siemens Skyra 3T
Showing 1-4 of 4 posts
Dec 9, 2015 03:12 AM | Victor Borges - University of Auckland
dcm2nii bvals from Siemens Skyra 3T
I'm hoping this actually reaches Chris Rorden.
New to acquire DWI data of a Siemens scanner.
Process the dicoms through dcm2nii in order to get the bval and bvecs for analysis. I've noticed that the bval are off. As in where the b-value should be zero it is 5 and where the b-value should be 2000 it can be 1995 or 2005 or 2010, but never 2000. Is this something to do with the multibanding or the way in which the Siemens acquire the dMRI?
Any information or suggestions would be most appreciated, cause I can't seem to find a simple explanation anywhere.
Cheers,
Victor
New to acquire DWI data of a Siemens scanner.
Process the dicoms through dcm2nii in order to get the bval and bvecs for analysis. I've noticed that the bval are off. As in where the b-value should be zero it is 5 and where the b-value should be 2000 it can be 1995 or 2005 or 2010, but never 2000. Is this something to do with the multibanding or the way in which the Siemens acquire the dMRI?
Any information or suggestions would be most appreciated, cause I can't seem to find a simple explanation anywhere.
Cheers,
Victor
Dec 15, 2015 01:12 AM | Victor Borges - University of Auckland
RE: dcm2nii bvals from Siemens Skyra 3T
I found this in a dcm2nii document
"Last, two additional notes:
1) The issue of bvec rotation is not strictly limited to oblique acquisitions. For example, if you collect a strictly coronal acquisition under Siemens VB13, VB15, or VB17, and you don't apply an angulation correction to the directions stored in the DICOM, then your ensuing bvecs will not be correctly specified relative to the imaging grid. [Specifically, the NIfTI file returned by dcm2nii will be oriented +LSP (left-superior-posterior), while the directions coded in the DICOM itself will follow a +LPS convention (consistent with the DICOM standard)]. Similarly, strictly sagittal acquisitions would require angulation correction (i.e., rotation) of the bvecs stored in the Siemen's DICOMs as well.
2) If you manually reorient the imaging grid following use of dcm2nii (e.g., using 'fslreorient2std'), then you also need to account for that reorienting on the bvecs as well."
But how do we correct this?? Cause I do have to use the fslreorient2std tool.
Also does anyone know how to get the bvecs form B0 image acquired separately from the main DWI?
Cheers,
Victor
Originally posted by Victor Borges:
"Last, two additional notes:
1) The issue of bvec rotation is not strictly limited to oblique acquisitions. For example, if you collect a strictly coronal acquisition under Siemens VB13, VB15, or VB17, and you don't apply an angulation correction to the directions stored in the DICOM, then your ensuing bvecs will not be correctly specified relative to the imaging grid. [Specifically, the NIfTI file returned by dcm2nii will be oriented +LSP (left-superior-posterior), while the directions coded in the DICOM itself will follow a +LPS convention (consistent with the DICOM standard)]. Similarly, strictly sagittal acquisitions would require angulation correction (i.e., rotation) of the bvecs stored in the Siemen's DICOMs as well.
2) If you manually reorient the imaging grid following use of dcm2nii (e.g., using 'fslreorient2std'), then you also need to account for that reorienting on the bvecs as well."
But how do we correct this?? Cause I do have to use the fslreorient2std tool.
Also does anyone know how to get the bvecs form B0 image acquired separately from the main DWI?
Cheers,
Victor
Originally posted by Victor Borges:
I'm hoping this actually reaches Chris
Rorden.
New to acquire DWI data of a Siemens scanner.
Process the dicoms through dcm2nii in order to get the bval and bvecs for analysis. I've noticed that the bval are off. As in where the b-value should be zero it is 5 and where the b-value should be 2000 it can be 1995 or 2005 or 2010, but never 2000. Is this something to do with the multibanding or the way in which the Siemens acquire the dMRI?
Any information or suggestions would be most appreciated, cause I can't seem to find a simple explanation anywhere.
Cheers,
Victor
New to acquire DWI data of a Siemens scanner.
Process the dicoms through dcm2nii in order to get the bval and bvecs for analysis. I've noticed that the bval are off. As in where the b-value should be zero it is 5 and where the b-value should be 2000 it can be 1995 or 2005 or 2010, but never 2000. Is this something to do with the multibanding or the way in which the Siemens acquire the dMRI?
Any information or suggestions would be most appreciated, cause I can't seem to find a simple explanation anywhere.
Cheers,
Victor
Dec 15, 2015 03:12 PM | Chris Rorden
RE: dcm2nii bvals from Siemens Skyra 3T
Victor-
A.) Note that when you setup a Siemens scanner you request gradients such as 0 and 1500. Based on this you might expect the applied gradients to only have these values, e.g.
0 0 0 0 1500 1500 1500 1500...
but in reality the scanner saves the applied gradients (with a bit of rounding error) rather than the requested gradients, so your file will look something like
5 5 5 5 1495 1500 1500 1490..
to quote Essa Yacoub these are "more accurate representations of the actual b values... Regarding the b zero - there are always gradients so the weighting is never zero but the crusher gradients typically amount to around 5 and we did add an option in the gui to increase that if need be".
Therefore, my software appropriately records the applied gradients (as the requested gradients are not stored in your images). Regardless, these are the most accurate values and the ones you will want to use in your post processing.
As an aside, as Paul Morgan notes "people generate B0 images with a low but non-zero b-value (e.g. 50) to suppress the CSF signal and therefore reduce partial volume effects for MD maps", which you may want to think about when you set up your sequence.
B.) By default the Siemens scanners typically only acquire one B0 value for each set of gradient directions, so if you have a 64-direction DTI scan Siemens will acquire 65 volumes (one B0, 64 with a gradient applied). This is potentially OK for FA, but for spatial unwarping and MD it is usually a good idea to have around 1/8 to 1/10th of your volumes to be B0. There are 3 ways to achieve this
1.) On the latest Siemens software (e.g. VD1x) the Siemens vendor sequences allow you to adjust the number of averages separately for each B value.
2.) Some people acquire a whole series of B0 images, in this case you can just create .bval and .bvec files with the appropriate number of zeros.
3.) The method I use (which works on Siemens prior to VD, and also with the fantastic CMRR sequences with VD) is to set up my own directions and add extra B0 images into the sequence. I outline this approach here
http://www.mccauslandcenter.sc.edu/CRNL/...
C.) The popular tools assume that gradient directions are reported in image space. If you reorient you images you simply apply the rotation matrix to the gradient directions:
https://en.wikipedia.org/wiki/Rotation_m...
You can check your solution using the method described in the Word document here
http://www.mccauslandcenter.sc.edu/mricro/mricron/dcm2nii.html
However, I would strongly discourage this approach. I strongly believe it is best to do all of your DTI pre-processing in native space. This is how the FSL tools were designed, and you may find all sorts of issues if you disobey this (e.g. as I recall tools like TOPUP make assumptions that the data is an even number of rows and columns but not slices).
A.) Note that when you setup a Siemens scanner you request gradients such as 0 and 1500. Based on this you might expect the applied gradients to only have these values, e.g.
0 0 0 0 1500 1500 1500 1500...
but in reality the scanner saves the applied gradients (with a bit of rounding error) rather than the requested gradients, so your file will look something like
5 5 5 5 1495 1500 1500 1490..
to quote Essa Yacoub these are "more accurate representations of the actual b values... Regarding the b zero - there are always gradients so the weighting is never zero but the crusher gradients typically amount to around 5 and we did add an option in the gui to increase that if need be".
Therefore, my software appropriately records the applied gradients (as the requested gradients are not stored in your images). Regardless, these are the most accurate values and the ones you will want to use in your post processing.
As an aside, as Paul Morgan notes "people generate B0 images with a low but non-zero b-value (e.g. 50) to suppress the CSF signal and therefore reduce partial volume effects for MD maps", which you may want to think about when you set up your sequence.
B.) By default the Siemens scanners typically only acquire one B0 value for each set of gradient directions, so if you have a 64-direction DTI scan Siemens will acquire 65 volumes (one B0, 64 with a gradient applied). This is potentially OK for FA, but for spatial unwarping and MD it is usually a good idea to have around 1/8 to 1/10th of your volumes to be B0. There are 3 ways to achieve this
1.) On the latest Siemens software (e.g. VD1x) the Siemens vendor sequences allow you to adjust the number of averages separately for each B value.
2.) Some people acquire a whole series of B0 images, in this case you can just create .bval and .bvec files with the appropriate number of zeros.
3.) The method I use (which works on Siemens prior to VD, and also with the fantastic CMRR sequences with VD) is to set up my own directions and add extra B0 images into the sequence. I outline this approach here
http://www.mccauslandcenter.sc.edu/CRNL/...
C.) The popular tools assume that gradient directions are reported in image space. If you reorient you images you simply apply the rotation matrix to the gradient directions:
https://en.wikipedia.org/wiki/Rotation_m...
You can check your solution using the method described in the Word document here
http://www.mccauslandcenter.sc.edu/mricro/mricron/dcm2nii.html
However, I would strongly discourage this approach. I strongly believe it is best to do all of your DTI pre-processing in native space. This is how the FSL tools were designed, and you may find all sorts of issues if you disobey this (e.g. as I recall tools like TOPUP make assumptions that the data is an even number of rows and columns but not slices).
Dec 15, 2015 09:12 PM | Victor Borges - University of Auckland
RE: dcm2nii bvals from Siemens Skyra 3T
Thanks Chris.
This has been super informative, still a bit over my head, but very helpful.
Thanks,
Victor
This has been super informative, still a bit over my head, but very helpful.
Thanks,
Victor