An Enhanced Registration and Display Algorithm for Medical Augmented Reality

摘要

Augmented Reality (AR), combines real world and supplemental aids, helps in better user perception which can be very useful inaccurate decision making especially in medical applications. AR process includes registration, display, and tracking. AR system must be of a very high accuracy and should have a clear standard way of saving metadata and various information for an AR standard display. A lot of registration techniques may lack in accuracy, and display techniques may not have a standard display on different browsers and platforms. This paper handles the above shortcomings by proposing a marker-less medical AR system that uses both of intensity based detection to detect lesions in the liver and augmented reality markup language (ARML) for a standard AR display. Morphological operations are applied to make detection easier and to enhance the scene. Detection will be then followed by AR registration technique which is implemented using a mutual based information algorithm. In all system phases including preprocessing, we will consider saving metadata and information in ARML tags. Finally tracking process will be considered by applying previous steps to consequent video frames. Results of the proposed system were found to enhance physician perception for targeted objects with high accuracy. In addition; the usage of ARML for saving metadata guarantees standard AR display, it also enhances patient treatment and follow up.