AUGMENTED REALITY FOR NEURONAVIGATION

Student thesis: Master's Thesis

Abstract

M icrosoft’s HoloLens, when released in 2016, was one of the first commercial headset devices to implement augmented reality on a commercial scale. It has since been the focus of discussion and research in the biomedical field as it allows for the potential marriage of the patient with their medical image data; a long overdue paradigm shift in Neuronavigation. The following work investigates the augmentation of HoloLens with data from its front-facing RGB camera to provide automatic registration of medical images onto their phantom target. Furthermore, it details a proper 3D control system for performing manual registration, demonstrates an implementation of the "keyhole-effect" for holographic data in order to achieve more practical experience, and looks at whether the current use of the spatial mapping data is of any benefit in practice. Subsequently, automatic registration using RGB camera information has shown to yield more consistent results than manual registration. Continuous automatic registration was able to maintain hologram registration with a mean perceived drift of 1.41mm, as well as a sub two-millimeter surface point localization accuracy of 87%, all while allowing the user to walk about a test area. This represents a 78% improvement for the later and first for the former given the current state of the art. Additionally it was shown that when using an active tracking schema, the use of constructing a spatial mesh is superfluous. Lastly, the successful windowing of holographic data has been seen as an advantage by those surgeons who have seen it.
Date of Award2019
Original languageEnglish
SupervisorJef Vandemeulebroucke (Promotor), Bart Jansen (Promotor) & Johnny Duerinck (Promotor)

Cite this

'