[Neuroimaging] 3d registration for data fusion in material science

Fri Aug 21 13:54:14 CEST 2015

The nireg repo was originally which push me to contact the Neuroimaging group.
Apart some scaling issue (anisotropic pixel spacing), the rigid body approach should suit to my problematic.

The following snippet :
https://github.com/nipy/nireg/blob/master/examples/affine_registration.py
seems a good starting points to me.

The magic is confine  into the HistogramRegistration class in the following module :
https://github.com/nipy/nipy/blob/master/nipy/algorithms/registration/histogram_registration.py
The code seems tricky, but with a lot of comments and docstrings !! it could have been  much much worse :)

Questions :

-          how to convert a numpy array to nipy-like image ?

-          should the two images have the same size ?

-          the version of nipy that I should use ?

thank you, really, for your attention, this is kind.

Gael

De : Neuroimaging [mailto:neuroimaging-bounces+gael.goret=cea.fr at python.org] De la part de Alexis Roche
Envoyé : vendredi 21 août 2015 13:04
À : Neuroimaging analysis in Python
Objet : Re: [Neuroimaging] 3d registration for data fusion in material science

Hi Gael,
Beside the stuff in dipy that Elef mentioned, there is another brain image registration package that has slowly developed over the years, and could be useful to you (although it still crucially lacks documentation):

https://github.com/nipy/nireg

This is rigid/affine registration only for the time being and has a BSD license too.

Best,
Alexis

On Fri, Aug 21, 2015 at 9:57 AM, GORET Gael 246279 <Gael.GORET at cea.fr<mailto:Gael.GORET at cea.fr>> wrote:
Hi all,

Thanks for your quick replies,
This project is just starting, and there is not too much materials …
However, to answer your questions, I just have created a public repo on my github account :

https://github.com/ggoret/MUDRA

-          In /doc, I have placed a summary (a pdf slideshow) of the project containing some pictures (and info on instruments)

-          In /examples I have put two data volumes (npy binary file format) the kind of data I need to register.

-          In /mudra a first (naive) try using the Fourier shell correlation as metric

And you can also find

-          a Cython wrapping of malik and perona’s algorithms implementation working (pretty quickly) on 3D Volumes :

o   /mudra/extensions/non_linear_filtering.pyx

-          In /scripts you will find mainly converters simplifying I/O

-          In /tools 3 very nice visualization tools based on VTK (not mayavi) working as standalone (npy input) :

o   elevation.py -> 2d image to 3d landscape

o   plan_interpolator.py -> 2d slicing of volume + isosurface rendering for given seeds

o   scalar_field.py -> volume rendering (color and opacity gradient)

Thanks in advance for your advices

Cheers,

Gael

De : Neuroimaging [mailto:neuroimaging-bounces+gael.goret<mailto:neuroimaging-bounces%2Bgael.goret>=cea.fr at python.org<mailto:cea.fr at python.org>] De la part de Eleftherios Garyfallidis
Envoyé : jeudi 20 août 2015 16:35
À : Neuroimaging analysis in Python
Objet : Re: [Neuroimaging] 3d registration for data fusion in material science

Hi Gael,

Sounds exciting. There is no restriction to use our tools in other libraries or domains.
Look at these tutorials please (you need dipy development version)

https://github.com/nipy/dipy/blob/master/doc/examples/affine_registration_3d.py
https://github.com/nipy/dipy/blob/master/doc/examples/syn_registration_2d.py
https://github.com/nipy/dipy/blob/master/doc/examples/syn_registration_3d.py

Do you have a github repo of your project? Do you have any example pictures/volumes to show us that you would
like to register?

Cheers,
Eleftherios

On Thu, Aug 20, 2015 at 9:31 AM, GORET Gael 246279 <Gael.GORET at cea.fr<mailto:Gael.GORET at cea.fr>> wrote:
Hi everybody,
I am a (French) newcomer, working on the development of the data fusion of Time-of-Flight Secondary Ion Mass Spectrometer (ToF-SIMS) and X ray Nano-Tomography (XuM) data at the CEA (the French state institute for energy, high-tech, etc.).
In terms of samples, we’re a bit far from the neuro-fields … Solid oxide fuel cell, 3D chips, Si/Li matrix, etc., but in terms of methodology I hope we have to share … my project is to combine a chemical information (from ToF-SIMS) with X-ray absorption (given by XuM) for a given volume (at nano-scale).  I am tackling a python module aiming to the registration of 3d datasets. it seems that nipy includes a such possibility (an a lot more …).
I’m a computer guy, mostly Pythonist and I wander if you would let  me transpose your code (mainly the registration part) to my problematic, I would be very grateful for this.
In this context would you have some advice for me ?
I'm looking forward to hearing from you.
Gael

[cid:image001.png at 01D0DC13.C3A9F850]

Dr. Gaël Goret
Chercheur Postdoctoral
Département des Technologies Silicium
Service de Caractérisation des Matériaux & Composants

Commissariat à l’énergie atomique et aux énergies alternatives
MINATEC Campus | 17 rue des martyrs | F-38054 Grenoble Cedex
T. +33 (0)4 38 78 49 29 | gael.goret at cea.fr<mailto:gael.goret at cea.fr>

_______________________________________________
Neuroimaging mailing list
Neuroimaging at python.org<mailto:Neuroimaging at python.org>
https://mail.python.org/mailman/listinfo/neuroimaging

_______________________________________________
Neuroimaging mailing list
Neuroimaging at python.org<mailto:Neuroimaging at python.org>
https://mail.python.org/mailman/listinfo/neuroimaging

--
Lead Clinical Research
Advanced Clinical Imaging Technology
Siemens/CHUV/EPFL
1015 Lausanne, Switzerland
Phone: +41 21 545 9972
https://sites.google.com/site/alexisroche
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/neuroimaging/attachments/20150821/12826288/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 8992 bytes
Desc: image001.png
URL: <http://mail.python.org/pipermail/neuroimaging/attachments/20150821/12826288/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.jpg
Type: image/jpeg
Size: 1321 bytes
Desc: image002.jpg
URL: <http://mail.python.org/pipermail/neuroimaging/attachments/20150821/12826288/attachment-0001.jpg>