Kakadiaris, Ioannis A.2019-09-142019-09-14May 20172017-05May 2017https://hdl.handle.net/10657/4567Augmented Reality can be useful in medical science because the visualization of a patient's internal structure is important for planning an intervention. Traditional display technologies present images collected from CT, MRI or ultrasound on 2D screens. However, aligning those images to a patient's body helps physicians decide the target of the intervention. In this thesis, an iPad application is proposed, which overlays 3D models of a patient's organs acquired using MRI scan in real-time video. The process is divided into two parts: 1) 3D registration and 2) tracking. The registration consists of acquiring the 3D structure of patient's torso using a depth sensor and overlaying pre-operative 3D models in real-time video. This is accomplished by a 3D point-to-point registration using the Iterative Closest Point algorithm. The tracking keeps the 3D models aligned on the patient's body by detecting the camera's pose. This is accomplished by using a technique called Simultaneous Localization and Mapping, which maps the environment and estimates the camera pose continuously. Registration accuracy is measured based on a study of eight different users of the application. The possibility of porting the application to a head-mounted display or glasses is also explored.application/pdfengThe author of this work is the copyright owner. UH Libraries and the Texas Digital Library have their permission to store and provide access to this work. Further transmission, reproduction, or presentation of this work is prohibited except with permission of the author(s).Medical visualizationAugmented realityDepth sensorImage registrationMedical imagingIterative Closest PointSimultaneous Localization and MappingCamera Pose2019-09-14Thesisborn digital