Automatic Segmentation of the Optic Radiation Using DTI in Healthy Subjects and Patients with Glaucoma

The complexity of the diffusion tensor imaging (DTI) data and the interpersonal variability of the brain fiber structure make the identification of the fibers a difficult and time consuming task. In this work, an automated segmentation system of the optic radiation using DTI is proposed. The system is applicable to normal subjects and glaucoma patients. It is intended to aid future glaucoma studies. The automation of the system is based on utilizing physiological and anatomical information to produce robust initial estimates of the optic radiation. The estimated optic radiation initializes a statistical level set framework. The optic radiation is segmented by the surface evolution of the level set function. The system is tested using eighteen DTI-datasets of glaucoma patients and normal subjects. The segmentation results were compared to the manual segmentation performed by a physician experienced in neuroimaging and found to be in agreement with the known anatomy with 83% accuracy. The automation eliminates the necessity of medical experts' intervention and facilitates studies with large number of subjects.