how_to
how_to > RE: HOW TO use the DTI template for registration
Jan 20, 2016 05:01 AM | Rowena C - UCL
RE: HOW TO use the DTI template for registration
Hi Konstantinos,
Just curious: will the registration not work if I did the steps but register to the 182 space template?
Best,
Rowena
Originally posted by Konstantinos Arfanakis:
Just curious: will the registration not work if I did the steps but register to the 182 space template?
Best,
Rowena
Originally posted by Konstantinos Arfanakis:
Hi Rowena,
I think the jhu labels are in 182 space (not 256). As a result, the output of DTI-TK using our 256 atlas will most likely not work with the jhu labels. I should mention here that you may want to take a look into the "HOW-TO-use-the-WM-atlas". The IIT white matter atlas contains four-dimensional (4D) white matter labels. This means that each white matter voxel is assigned a list of gray matter label-pairs with the most probable connections through that voxel. The reason behind the 4D labels is that multiple fiber populations often share the same voxels, and therefore white matter voxels must be given the chance to have multiple labels. If you still prefer the winner-takes-all approach (similar to the jhu approach) then you can simply use the top label in each voxel of our atlas, which means you can simply use the first volume of our 4D atlas and ignore the rest of the volumes (less likely labels). However, I recommend avoiding "winner-takes-all" white matter atlases, for the reason explained above. It is best to embrace the truth i.e. multiple connections can share the same white matter voxels, so each white matter voxel should have multiple labels.
Please let me know if you have more questions.
Regards,
Konstantinos
I think the jhu labels are in 182 space (not 256). As a result, the output of DTI-TK using our 256 atlas will most likely not work with the jhu labels. I should mention here that you may want to take a look into the "HOW-TO-use-the-WM-atlas". The IIT white matter atlas contains four-dimensional (4D) white matter labels. This means that each white matter voxel is assigned a list of gray matter label-pairs with the most probable connections through that voxel. The reason behind the 4D labels is that multiple fiber populations often share the same voxels, and therefore white matter voxels must be given the chance to have multiple labels. If you still prefer the winner-takes-all approach (similar to the jhu approach) then you can simply use the top label in each voxel of our atlas, which means you can simply use the first volume of our 4D atlas and ignore the rest of the volumes (less likely labels). However, I recommend avoiding "winner-takes-all" white matter atlases, for the reason explained above. It is best to embrace the truth i.e. multiple connections can share the same white matter voxels, so each white matter voxel should have multiple labels.
Please let me know if you have more questions.
Regards,
Konstantinos
Threaded View
Title | Author | Date |
---|---|---|
Konstantinos Arfanakis | Jun 6, 2013 | |
magan | Feb 13, 2018 | |
Konstantinos Arfanakis | Feb 14, 2018 | |
magan | Feb 15, 2018 | |
Rowena C | Oct 16, 2015 | |
Konstantinos Arfanakis | Oct 16, 2015 | |
Rowena Chin | Oct 21, 2015 | |
Konstantinos Arfanakis | Oct 22, 2015 | |
Rowena C | Jan 20, 2016 | |
Konstantinos Arfanakis | Jan 21, 2016 | |
Lauren Byrne | Mar 31, 2015 | |
Lauren Byrne | Mar 26, 2015 | |
Konstantinos Arfanakis | Mar 30, 2015 | |
Lauren Byrne | Mar 31, 2015 | |
Lauren Byrne | Mar 31, 2015 | |
Konstantinos Arfanakis | Apr 3, 2015 | |
Lauren Byrne | Apr 10, 2015 | |
Konstantinos Arfanakis | Apr 10, 2015 | |
Lauren Byrne | Apr 14, 2015 | |
Konstantinos Arfanakis | Apr 22, 2015 | |
Lauren Byrne | May 11, 2015 | |
Konstantinos Arfanakis | May 13, 2015 | |
Lauren Byrne | May 14, 2015 | |
Konstantinos Arfanakis | May 14, 2015 | |
Lauren Byrne | May 15, 2015 | |
Lauren Byrne | May 11, 2015 | |
Lauren Byrne | Apr 24, 2015 | |
Marcus heldmann | May 27, 2014 | |
Konstantinos Arfanakis | May 28, 2014 | |
Konstantinos Arfanakis | Aug 10, 2013 | |
Sean Hatton | Jan 2, 2014 | |
Konstantinos Arfanakis | Jan 3, 2014 | |