Development of an Atlas-Based Segmentation of Cranial Nerves Using Shape-Aware Discrete Deformable Models for Neurosurgical Planning and Simulation
Date of Award
Doctor of Philosophy (PhD)
Computational Modeling & Simulation Engineering
Modeling and Simulation
Michel A. Audette
Twelve pairs of cranial nerves arise from the brain or brainstem and control our sensory functions such as vision, hearing, smell and taste as well as several motor functions to the head and neck including facial expressions and eye movement. Often, these cranial nerves are difficult to detect in MRI data, and thus represent problems in neurosurgery planning and simulation, due to their thin anatomical structure, in the face of low imaging resolution as well as image artifacts. As a result, they may be at risk in neurosurgical procedures around the skull base, which might have dire consequences such as the loss of eyesight or hearing and facial paralysis. Consequently, it is of great importance to clearly delineate cranial nerves in medical images for avoidance in the planning of neurosurgical procedures and for targeting in the treatment of cranial nerve disorders. In this research, we propose to develop a digital atlas methodology that will be used to segment the cranial nerves from patient image data. The atlas will be created from high-resolution MRI data based on a discrete deformable contour model called 1-Simplex mesh. Each of the cranial nerves will be modeled using its centerline and radius information where the centerline is estimated in a semi-automatic approach by finding a shortest path between two user-defined end points. The cranial nerve atlas is then made more robust by integrating a Statistical Shape Model so that the atlas can identify and segment nerves from images characterized by artifacts or low resolution. To the best of our knowledge, no such digital atlas methodology exists for segmenting nerves cranial nerves from MRI data. Therefore, our proposed system has important benefits to the neurosurgical community.
"Development of an Atlas-Based Segmentation of Cranial Nerves Using Shape-Aware Discrete Deformable Models for Neurosurgical Planning and Simulation"
(2017). Doctor of Philosophy (PhD), Dissertation, Computational Modeling & Simulation Engineering, Old Dominion University, DOI: 10.25777/m30g-bw21