Storage Optimization for an Immersive and Collaborative Scientific Visualization System of 3D Objects

Abstract

This paper presents a novel system for storage, administration, and immersive photo-realistic visualization of volumetric scenes for interdisciplinary scientific uses, specifically applied to the fields of academic medical training and museology. Using virtual reality, the system creates a collaborative work environment that allows for the visualization and analysis of 3D objects stored in two different formats. Objects with interior and exterior information are represented with volumes composed of voxels. Objects with only surface information are represented by textured polygonal meshes. While computer tomography (CT) or magnetic resonance imaging (MRI) scanners are utilized to generate 3D volumes, polygonal meshes are obtained by scanners, photogrammetric methods, and more recently, through 3D reconstruction technologies based on Artificial Intelligence called Neural Radiation Fields (NeRF), which present advantages given the visual results obtained. The 3D objects generated from either of the two alternative formats require high-capacity storage, which still represents a problem to be solved. This work proposes the use of NeRF, not only as a reconstruction technology, but also as a more efficient method for storing data, at the expense of performing a real-time conversion to polygonal meshes for visualization.