[Neurobureau-hubs] computational resources

[Pierre Bellec]

Hi hubs,

I am forwarding to you an email from Romain Valabregue re brainhack and the
possibility to access local computational resources.

Hello Dear NeuroBureau

I am a research engineer in charge the CENIR data acquisition plateform
at the ICM (paris).
Since I am also in charge of the computing resources (containing a small
cluster of 400 nodes + 2 big openmp1 machine torque manager, nodes with
centos6 ). I could also give access to several single server for
individual computing.
The access could be donne for any labtop with the nxclient
(http://www.nomachine.com/download.php)

Tel me if you will be interested so that I can think of the best way to
achieve it. (my first idea was to give only a single account.)


About data Resources I already download the DTI and resting state data
from the connectome project. So It will be
available. (40 sujbects HCP Q2 Release :( a very nice preprocessed High
resolution data set) http://humanconnectome.org/data/) there are also
task fMRI data that I could download if needed



I am very interested to come to brainhack, but I hesitate to posted a
project since I do not see any proposal.
there was several project in 2012 (C-PAC, fmri quality control, xnat
...) that are of primary interest for my propose,
but I do not see neither what do they achieve nor if they will be
present in 2013. Is it normal to do not see any project for this year ?

In short, my project idea was to propose 3 to 4 years of scaning raw
data of our 3 T trio and Verio scanner and address the question of
quality control in two directions : what are the bad data and what are
the best data :
1_ Give quality indices of each data set and automatically detect
artefact  (spykes, motions, background noise...)
2_ given the variability of acquisitions which parameters are the best
(in term of SNR contrast ratio connectivity ...)
There is also a dimension of modality the procedure may be adapt for
fMRI DWI and anatomical scan

There is a need of quality indices, a minimum of databasing to take care
the variability of acquisitions, both integrated in a automatic
framework that distribute job to the cluster resources. It seems to me
that there are several solutions that are accessible today (even just
looking at 2012 brainhack projects).

I only propose the data and the computing resources and my hope is that
expert of those available solutions will take the challenge in the short
time of the conference to process our 10 thousand exams. But I am not
sure if this is a viable project and if they will find enough interest
(time) to give a try...



Cheers

Romain Valabregue

CENIR, ICM
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/neurobureau-hubs/attachments/20131008/bd46193d/attachment.html