Parallel Hierarchical Agglomerative Clustering for fMRI Data
This paper describes three parallel strategies for Ward’s algorithm with OpenMP or/and CUDA. Faced with the difficulty of a priori modelling of elicited brain responses by a complex paradigm in fMRI experiments, data-driven analysis have been extensively applied to fMRI data. A promising approach is clustering data which does not make stringent assumptions such as spatial independence of sources. Thirion et al. have shown that hierarchical agglomerative clustering (HAC) with Ward’s minimum variance criterion is a method of choice. However, HAC is computationally demanding, especially for distance computation. With our strategy, for single subject analysis, a speed-up of up to 7 was achieved on a workstation. For group analysis (concatenation of several subjects), a speed-up of up to 20 was achieved on a workstation.
KeywordsHierarchical agglomerative clustering OpenMP CUDA fMRI Distance computation
- 3.Chang, D., et al.: Compute pairwise Euclidean distances of data points with GPUs. In: Proceedings of the IASTED International Symposium Computational Biology and Bioinformatics (CBB 2008) (2008)Google Scholar
- 4.Cordes, D., Haughton, V.M., Arfanakis, K., et al.: Mapping functionally related regions of brain with functional connectivity MR imaging. Am. J. Neuroradiol. 21, 1636–1644 (2000)Google Scholar
- 7.Gao, X., et al.: Comparison between spatial and temporal independent component analysis for blind source separation in fMRI data. In: 2011 4th International Conference on Biomedical Engineering and Informatics (BMEI), vol. 2, pp. 690–692. IEEE (2011)Google Scholar
- 10.Kim, S., Ouyang, M.: Compute distance matrices with GPU. Glob. Sci. Technol. Forum (2012). https://doi.org/10.5176/2251-1652_ADPC12.07
- 12.Li, Q., et al.: A chunking method for Euclidean distance matrix calculation on large dataset using multi-GPU, pp. 208–213. IEEE (2010)Google Scholar