Illustrative white matter fiber bundles

Research output: Contribution to journalArticleAcademicpeer-review

15 Citations (Scopus)


Diffusion Tensor Imaging (DTI) has made feasible the visualization of the fibrous structure of the brain whitematter. In the last decades, several fiber-tracking methods have been developed to reconstruct the fiber tracts fromDTI data. Usually these fiber tracts are shown individually based on some selection criteria like region of interest.However, if the white matter as a whole is being visualized clutter is generated by directly rendering the individualfiber tracts. Often users are actually interested in fiber bundles, anatomically meaningful entities that abstractfrom the fibers they contain. Several clustering techniques have been developed that try to group the fiber tractsin fiber bundles. However, even if clustering succeeds, the complex nature of white matter still makes it difficultto investigate. In this paper, we propose the use of illustration techniques to ease the exploration of white matterclusters. We create a technique to visualize an individual cluster as a whole. The amount of fibers visualized forthe cluster is reduced to just a few hint lines, and silhouette and contours are used to improve the definition of thecluster borders. Multiple clusters can be easily visualized by a combination of the single cluster visualizations.Focus+context concepts are used to extend the multiple-cluster renderings. Exploded views ease the explorationof the focus cluster while keeping the context clusters in an abstract form. Real-time results are achieved by theGPU implementation of the presented techniques. Keywords: I.3.3 [Computer Graphics]: Picture/Image Generation—Bitmap and framebuffer operations, Display algorithms, Viewing algorithms.
Original languageEnglish
Pages (from-to)1013-1022
JournalComputer Graphics Forum
Issue number3
Publication statusPublished - 2010


Dive into the research topics of 'Illustrative white matter fiber bundles'. Together they form a unique fingerprint.

Cite this