-
Notifications
You must be signed in to change notification settings - Fork 429
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[ENH] Compute fiber density and fiber spread from ODF using Bingham distributions #2826
base: master
Are you sure you want to change the base?
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #2826 +/- ##
==========================================
- Coverage 83.76% 83.60% -0.16%
==========================================
Files 153 153
Lines 21340 21468 +128
Branches 3445 3455 +10
==========================================
+ Hits 17876 17949 +73
- Misses 2608 2646 +38
- Partials 856 873 +17
|
Hi @CHrlS98, thanks for initiating this pull request. Regarding your questions:
|
Hello @CHrlS98, Thank you for updating ! Cheers ! There are no PEP8 issues in this Pull Request. 🍻 Comment last updated at 2024-05-13 13:49:54 UTC |
1419292
to
ca6268a
Compare
Hello @CHrlS98, What is the status of this work? Do you know when you will have time to progress on this? ps: next release should be on beginning of March, it would be great to have this work |
Hi @skoudoro , |
5935e1e
to
765963e
Compare
…ion to fit bingham distribution to a full brain image; add number of peaks to the fitting; change eig to eigh; modify documentation.
…ation to example in bingham_fit_fodf.py. Comply with PEP8 standard.
Hi @villalonreina, No idea, but when it does not show up, just go directly to his repo and it will show up: https://github.com/CHrlS98/dipy/pulls |
Odf to bingham
@villalonreina @RafaelNH
|
Notes from our last conversation. To do:
|
Hi @RafaelNH, is it possible t-design is slightly biased? I create my sphere this way (it doesn't matter if I subdivide it or not, I tried both) :
And then I added a test where I check if the integral of a Bingham distribution along the y-axis is equal to the integral along the x-axis:
And sure it's close, but not even that close:
However when I use a unit icosahedron and subdivide it 5 times (10242 vertices), the test passes. I'll keep the unit icosahedron for now because it is better distributed than repulsion spheres, and add an option to control the level of subdivision. Is the t-design a whole sphere or a hemisphere? Could it be the reason for this bias? |
|
||
# Test Bingham fit on full sampled GT Bingham function | ||
odf_gt = _single_bingham_to_sf(f0, k1, k2, ma_axis, mi_axis, sphere.vertices) | ||
a0, c1, c2, mu0, mu1, mu2 = _bingham_fit_peak(odf_gt, peak_dir, sphere, 45) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@RafaelNH If I replace peak_dir
by -peak_dir
I can lower the max angle to lesser than 45 and the test comparing Mus
and Mus_ref` will still pass.
As discussed with @villalonreina @RafaelNH @mdesco,
This is an attempt at migrating code from
scilpy
for fitting Bingham distributions to ODF and computing fiber metrics such as the fiber density and fiber spread [1]. The idea was that this could serve as a basis for GSoC 2023 Project 2: Generalized along tract analysis of fiber orientation dispersion.The PR is WIP. It still misses unit tests and stuff. I would mostly like some feedback at this point:
scilpy
script here. Of course it would be cool, but maybe not needed in a first PR.reconst
module. It's more of a voxel based analysis or something. Should this go in another module? Is it worth creating a new one?Checklist
References
[1] Riffert TW, Schreiber J, Anwander A, Knösche TR. Beyond fractional anisotropy: Extraction of bundle-specific structural metrics from crossing fiber models. NeuroImage. 2014 Oct 15;100:176-91.