An Interactive Algorithm for Virtual Patient Positioning

Abstract

In the last years, Virtual Reality medical simulators are gaining importance, training new physicians in a safe environment. In order to improve the trainees' skills, these applications let them perform a specific medical procedure in different virtual patients with different anatomical variation. Typically, virtual patients are generated from medical imaging data sets, such as MRI (magnetic resonance imaging), CT (computed tomography) or US (ultra sound). Usually, these data are generally captured in specific subject position. This pose is different from the subjects' position required in the simulated specific medical procedure. This paper proposes a novel technique that allows adapting the virtual patient anatomy to any desired pose. Our algorithm follows a geometrically based approach with the intention of: (i) being independent of a proper mechanical description of the tissues which is rarely available and (ii) keeping our user interface running at interactive rates. We adapted the skeletal animation workflow to deal with internal anatomy models. Most of the stages of this workflow have been automated. The user intervention has been limited to the interactive pose selection process. Additionally, in order to refine the solutions provided by our geometric approach, we have designed an optimization phase to achieve more appealing results.