PRASAD: An Augmented Reality based Non-invasive Pre-operative Visualization Framework for Lungs Anand P Santhanam (1), Cali Fidopiastis (2), Bari Hoffman-Ruddy, (3), Jannick P Rolland (1,4) (1) Department of Computer Science, University of Central Florida (2) Institute of Simulation and Training, University of Central Florida (3) Department of Health and Public Affairs (4) School of Optics/CREOL/FPCE, University of Central Florida anand@odalab.ucf.edu , cali@odalab.ucf.edu , bari@mail.ucf.edu , Jannick@odalab.ucf.edu Abstract This paper presents a pre-operative anatomical visualization framework, PRASAD (Physically Realistic Adaptive and Scalable Anatomical Deformation system), which combines a bio-mathematical representation of deformable lungs with real-time deformation and stereoscopic visualization technology. This framework provides a visualization of a dynamic patient-specific deformation of synthetic 3D anatomical models, that physicians can view from different viewpoints in a stereoscopic augmented reality environment for efficient diagnosis. 1. Introduction Creating a visualization of patient-specific anatomical models that displays physically and physiologically realistic dynamic deformations constitutes a significant contribution to medical training and teaching, and may also provide aids to diagnosis in clinical settings. 2. Related Work 2.1 Pulmonary Preoperative Assessment Surgery simulation has found a place in medical training centers as a means of allowing surgeons to visualize and rehearse surgical procedures in a safe and realistic environment.[1] Preoperative assessment protocols usually consist of a decision tree design where imaging techniques along with physiology measures are used to predict patient risk for developing pulmonary complications.[2] Of the assessment battery, pulmonary function has been the most studied. These techniques do not combine anatomical and physiological measures to give a complete pulmonary assessment. An anatomically correct 3D computer generated lung model that incorporates patient specific data with physiological measures of pressure-volume changes could bridge this gap. 2.2 Physiological Considerations for Lung Deformations First, the P-V curve obtained from physiology is used as a driver to the deformable model. Given a P-V curve, the physiology of alveoli distribution is used to specify the local shape of deformation. Specifically, in the normal at rest individual in the upright position, there is a natural intra-pleural pressure gradient from the upper lung region to the lower. The negative intra- pleural pressure at the apex of the lung is normally greater than at the base.[3] Thus the compliance of the alveoli in the upper lung region is normally less than the compliance of the alveoli in the lower lung regions in a normal person in the upright position. 3 Real-time Deformable Models for 3D Augmented Environments The main objective of the approach developed is to visualize real-time lung physiology-based deformations. We shall provide in section 4.1 the functional architecture of the framework. In section 4.2, we detail the approach to elastic lung deformations. In section 4.3, we discuss the bio-mathematical driver. In section 4.4, we discuss the hardware implementation, and the visualization system is provided in section 4.5. 4. PRASAD implementation for Lungs 4.1 Functional Architecture A detailed functional diagram of the software implementation is given in fig 1. The implementation is done using C++ in a Linux operating system. We consider two different model formats. A knowledge- base module decides the initial assignment for each node based on the bio-mathematical model of lungs.[4] The 3D Model Deformer module, updates the position IEEE Virtual Reality 2004 March 27-31, Chicago, IL USA 0-7803-8415-6/04/$20.00©2004 IEEE. 253 Proceedings of the 2004 Virtual Reality (VR’04) 1087-8270/04 $ 20.00 IEEE