Fang-Cheng
Yeh1, Wen-Yih Isaac Tseng2
1Department of Biomedical Engineering,
Carnegie Mellon University, Pittsburgh, PA, United States; 2Center
for Optoelectronic Biomedicine, National Taiwan University College of
Medicine, Taipei, Taiwan
We
present a tractography segmentation algorithm called morphological continuity
clustering (MCC), which is a fully automatic, unguided method that clusters
fiber tracts without predefining the cluster number. This algorithm is based
on the concept that the fibers of the same cluster share the morphological
continuity, a feature used to determine whether two tracts should be grouped.
The performance was evaluated on tractography with a total of 100,000 fibers
tracts generated by streamline tracking method on generalized q-space imaging
(GQI). The results showed that MCC is able to generate several clusters that
correspond to well-known fiber tracts. Further study is needed to improve the
accuracy and robustness of the proposed method.