White Matter Fiber Set Simplification by Redundancy Reduction with

Minimum Topological Information Loss



Diffusion Weighted Imaging (DWI) characterizes the diffusion of water in the tissues and is sensitive to the microstructural density, spacing, and orientational organization of tissue membranes. Fiber tracking or tractography exploits the measured orientation distribution of water diffusion to follow specific white matter pathways from voxel to voxel through the brain. This information is useful when studying the organization of white matter in the brain as well as the microstructural changes that occur with neuropathology and treatment. Advanced acquisition techniques, such as High Angular Resolution Imaging (HARDI) in conjunction with appropriate tractography algorithms produce highly dense fiber datasets, which can hold millions of fibers. Working with such huge datasets can be quite challenging, especially for algorithms that are supposed to perform highly complex operations on the data. Hence, one can ease the analysis of brain fibers by exploiting the redundancies of the sets, namely the presence of almost identical fibers. This is achieved by compressing the overall fiber set and keeping only unique representatives. Evaluating the similarity and distance between fibers is a key component in the process of analayzing the human brain and locating the main fibers which build up the main anatomical structures. Therefore, in our work we have compared various state-of-the-art distance metrics that have been used with WM fibers and developed a novel approach for comparing the metrics and estimating the topological information loss when keeping only a reduced set of fibers. Two conceptually different approaches to fiber set reduction have been tested: The first is deterministic and based on a series of clustering steps with a variety of distance metrics that have been commonly used in on fibers. The other is randomized and stems from the concept of Coresets. The latter was never applied to white matter fibers. Corsets presents a new approach to optimization and has huge success especially in tasks which use prohibitively large computation time and/or memory space, hence has the poential to be very beneficial for our purpose.




Zimmerman Moreno, G., Alexandroni, G. and Greenspan, H.: White Matter Fiber Set Simplification by Redundancy Reduction with Minimum Anatomical Information Loss.

MICCAI Workshop on Computational Diffusion MRI (2015).

horizontal rule

For problems or questions regarding this web page contact hayit@eng.tau.ac.il.
Last updated: 19/11/15.