Automatic Segmentation of the Human Hippocampus Using Superquadric Surface Model
Lee J, Hou Y, Wee W, Leach J, Cai X, Peng Z, Ball W
University of Cincinnati, University of Cincinnati
A new approach is proposed for the segmentation of the hippocampus. This approach is based on a statistical superquadric surface model and consists of two parts: the construction of a statistical hippocampal surface model and using this model to segment the hippocampus. The segmentation involves three steps: 1) Defining the region of interest that includes the hippocampus; 2) Classifying the grey matter (GM) within the region of interest as the GM initial condition; 3) Applying the parametric surface model to the GM initial condition and segment out the hippocampus. Finally, a local fine-tuning measure is applied to the segmentation result.