OpenCL Accelerated Connectome Analysis in Python
-
1
Max Planck Institute for Human Cognitive and Brain Sciences, Germany
In the last years there has been an increasing interest in the analysis of human connectome data. Connectivity is naturally at least bivariate since it is computed between each pair of objects. The computational complexity of connectome analysis grows therefore at least quadradic in resolution size. For example, in functional magentic resonance imaging (fMRI) resolution is increasing; not only with higher field strength but also with improved scanner sequences.
To cope with this increasing computational demand researchers have begun to apply two strategies. One strategy uses parcellation techniques to reduce dimensionality. The second approach are simpler and less demanding algorithms. However, both procedures encounter limitations. Here, we suggest an additional approach which concentrates on massive parallelization by exploiting modern and future hardware capabilities. Especially in the context of neuroimaging, many computational problems are easy to parallelize. So far, many parallel programming environments were to hardware specific to be beneficial for a wider user community. In this context we chose the emerging hardware independent OpenCL interface.
One simple and widely applied measure of functional connectivity is the Pearson product-moment correlation between hemodynamic signals. Even so Pearson correlation is relatively fast estimated, the computation on the whole-brain in a high resolution is computationally demanding. Here we show computing connectivity can be accelerated using parallelization in pyOpenCL (Fig. 1). Future GPGPU environments with thousands of cores and fast accessible memory will enable the analysis of larger datasets with even higher throughputs.
To foster brain research we want to further implement standard connectome analysis techniques in an easy accessible (python), hardware independent and well scalable (OpenCL) way.
Keywords:
GPU computing,
OpenCL,
connectome,
GPGPU,
Parallel Computing
Conference:
Neuroinformatics 2013, Stockholm, Sweden, 27 Aug - 29 Aug, 2013.
Presentation Type:
Poster
Topic:
Neuroimaging
Citation:
Schäfer
A and
Hellrung
L
(2013). OpenCL Accelerated Connectome Analysis in Python.
Front. Neuroinform.
Conference Abstract:
Neuroinformatics 2013.
doi: 10.3389/conf.fninf.2013.09.00061
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
29 Apr 2013;
Published Online:
11 Jul 2013.
*
Correspondence:
Mr. Alexander Schäfer, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, -- SELECT --, 04103, Germany, alexschaefer83@gmail.com