Software Framework for Middle Ear Optical Coherence Tomography

Abstract

Originally developed for ophthalmic imaging, optical coherence tomography (OCT), a non-invasive depth-resolved optical imaging modality, has recently been successfully applied to imaging of the middle ear through the intact eardrum. Middle ear OCT (ME-OCT) offers real-time structural and functional images of the middle ear health at point of care safely and without exposing the patient to ionizing radiation. As the use of ME-OCT technology evolves from feasibility studies to clinical application there is a need for system-level software engineering to integrate it into clinical workflow to enable clinicians to image and visualize diagnostic information independently and in a manner that can keep up with clinical workflow. This thesis introduces the design, implementation, verification, and validation of a flexible software framework meeting the performance and usability requirements for a ME-OCT imaging system suitable for real-world, clinical deployment. The framework provides a custom-built rendering engine capable of volumetric rendering of ME-OCT datasets at real-time interactive rates. Additionally, the rendering engine introduces two novel rendering techniques, digital tympanotomy for the real-time removal of the ear drum to visualize the underlying middle ear structures, and Doppler animation for intuitive visualization of the middle ear’s acoustic response to applied stimuli. This rendering engine is integrated into a purpose-built software architecture providing a turn-key imaging experience for integration into the clinical workflow without requiring close engineering support. The architecture was extended to incorporate two novel diagnostic imaging modalities: geometrically accurate, live, continuous volumetric ME-OCT imaging, and real-time ME-OCT angiography (ME-OCTA) for the direct, non-invasive, visualization of depth-resolved middle ear vasculature and dynamics in-vivo. Residual geometrical error was assessed by co-registration corrected ME-OCT and micro-CT datasets from both a 3D printed imaging phantom and cadaveric temporal bone. While in-vivo continuous volumetric imaging was demonstrated on a healthy adult volunteer during a dynamic pressurization maneuver. Real-time, phase-sensitive, 2D and 3D ME-OCTA in-vivo imaging was demonstrated for the first time and applied to visualization of middle ear vasculature and the stapedius reflex. Beyond ME-OCT, this research could be adapted for robotic intra-surgical applications, including surgical planning and real-time navigation, while also enabling post-surgery disease progression monitoring.