Method for the Automatic Segmentation of the Facial Nerve and the Chorda Tympani in CT Images


A system and method for automatic segmentation of a structure of interest of an ear are disclosed, where the structure of interest includes the facial nerve and chorda tympani. In one embodiment, the method combines an atlas-based approach with a minimum cost path finding algorithm. A structure model is constructed from a plurality of preoperative image volumes to have a centerline of the structure of interest in an atlas with each point along the centerline being associated with expected values for characteristic features, and used to create a spatially varying cost function that includes geometric information. A three-dimension minimum cost accumulating path is computed using the cost function, which is used to extract the centerline of the structure of interest. The centerlines are then expanded into the full structures using a level-set algorithm with a spatially-varying speed function, so as to segment of the structure of interest.

Intellectual Property Status

U.S. Patent # 8,073,216 was issued on December 6, 2011.

Licensing Contact

Taylor Jordan

Tech ID:
Medical Imaging