Qualitative, Quantitative, and Autonomous Optical Coherence Tomography Guided Ophthalmic Microsurgery
Access is limited until:
Ophthalmic microsurgery is a challenging subspecialty mainly due to the size of the operating environment. Tissues and distances are measured in micrometers and a surgical microscope is required to view the surgical field. Even with the microscope, surgeons have trouble visualizing all aspects of surgery because of the limited depth perception the microscope provides. Surgeons spend years developing the fine motor skills necessary to operate, but still encounter difficulty when performing certain procedures.
Recent developments in optical coherence tomography (OCT) have improved ophthalmic surgeons' capability to visualize surgery. OCT is a non-contact volumetric imaging modality capable of penetrating 1-2mm in tissue and is ideally suited for imaging the cornea and retina. Using OCT alongside the standard microscope view during surgery provides surgeons with more complete depth information. Despite improved visualization, movement and manipulation challenges in microsurgery persist, and surgeons have attempted to solve these problems through the use of robots. Robots offer improved accuracy and tremor reduction when positioning instruments and have been specifically developed for ophthalmic surgery.
This dissertation presents technologies to help aid surgeons in ophthalmic microsurgery. We begin by describing software capable of acquiring and processing real-time volumetric OCT to enhance the surgeon's view of the surgical field. Next, we report on methods for extracting quantitative information from OCT scans via retinal segmentation, corneal segmentation, and three-dimensional needle tracking. Finally, we combine OCT and robotics to make progress toward automating ophthalmic microsurgery. We use quantitative information from corneal segmentation and needle tracking together with reinforcement learning to enable a robot to teach itself to perform needle insertions in ex vivo tissue.
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Duke Dissertations