Optimization of RecoBundles for recognition of short white matter bundles
Résumé
Short-range connections have been less studied than long-range connectionq, with most of the white matter fibers atlases available containing mainly large connections. Among the short-range connections atlas, two developed by [6] managed to extract more than 400 bundles for one altas ( ARCHI database [8] ) and more than 600 bundles ( HCP database [1] ). However, they have not been used for white matter segmentation because most techniques for white matter segmentation are optimized for long range connections. Here, we optimize the technique implemented in [2] by analysing the atlas via six geometrical parameters and using bundle specific parameters to increase the quality and the number connections extracted.
Origine : Fichiers produits par l'(les) auteur(s)