Improvement of Simplex Meshes Model for 3D Hippocampus Segmentation

M.M. Karimi, N. Batmanghelich, H. Soltanian-Zadeh, and C. Lucas (Iran)


Simplex meshes, Deformable surface, brain segmentation, Magnetic Resonance Imaging (MRI)


In this paper, we used a deformable surface model, called simplex meshes, for hippocampus segmentation in brain MRI. Major problems of the hippocampus segmentation are weak edges and noise that may cause deformable model to move in wrong ways. To overcome these problems, we used simplex meshes model, which has the capability to move roughly. To initialize the primary shape we projected an atlas to the real data using a registration algorithm. We selected some parts of the initial shape and exerted forces to the vertices of this shape, which is proportion to the distance to these parts. Displacement of these parts helps model to overcome weak edges and also prevents self-cutting of the vertices. Finally, we did a final tuning to reach to the small details of the edges.

Important Links:

Go Back