Multi-Fiber Reconstruction from DW-MRI using a Continuous Mixture of von Mises-Fisher Distributions

Category:

Description

In this paper we propose a method for reconstructing the Diffusion Weighted Magnetic Resonance (DW-MR) signal at each lattice point using a novel continuousmixture of von Mises-Fisher distribution functions. Unlike most existing methods, neither does this model assume a fixed functional form for the MR signal attenuation (e.g. 2nd or 4th order tensor) nor does it arbitrarily fix important mixture parameters like the number of components. We show that this continuous mixture has a closed form expression and leads to a linear system which can be easily solved. Through extensive experimentation with synthetic data we show that this technique outperforms various other state-of-the-art techniques in resolving fiber crossings. Finally, we demonstrate the effectiveness of this method using real DW-MRI data from rat brain and optic chiasm.

Additional information

Author

Kumar, R., Barmpoutis, A., Vemuri, B. C., Carney, P. R., Mareci, T. H.

Journal

In Proceedings of MMBIA08: IEEE Computer Society Workshop on Mathematical Methods in Biomedical Image Analysis

Year

2008

Pages

1-8

DOI

https://doi.org/10.1109/CVPRW.2008.4562991

Citation

Citation

Kumar, R., Barmpoutis, A., Vemuri, B., Carney, P. and Mareci, T., 2008. Multi-Fiber Reconstruction from DW-MRI using a Continuous Mixture of von Mises-Fisher Distributions. In Proceedings of MMBIA08: IEEE Computer Society Workshop on Mathematical Methods in Biomedical Image Analysis, pp. 1-8. https://doi.org/10.1109/CVPRW.2008.4562991

BibTex

@article{digitalWorlds:167,
doi = {https://doi.org/10.1109/CVPRW.2008.4562991},
author = {Kumar, R. and Barmpoutis, A. and Vemuri, B. C. and Carney, P. R. and Mareci, T. H.},
title = {Multi-Fiber Reconstruction from DW-MRI using a Continuous Mixture of von Mises-Fisher Distributions},
journal = {In Proceedings of MMBIA08: IEEE Computer Society Workshop on Mathematical Methods in Biomedical Image Analysis},
year = {2008},
pages = {1-8}
}