<title>Real-time auto-stereoscopic visualization of 3D medical images</title>

Abstract

The work here described regards multi-viewer auto-stereoscopic visualization of 3D models of anatomical structures and organs of the human body. High-quality 3D models of more than 1600 anatomical structures have been reconstructed using the Visualization Toolkit (vtk), a freely available C++ class library for 3D graphics and visualization. 2D images used for 3D reconstruction come from the Visible Human Data Set (original and segmented images). Auto-stereoscopic threedimensional image visualization is obtained using a prototype monitor developed at Philips Research Labs, UK. This special multiview 3D-LCD screen has been connected directly to a SC}I workstation, where 3D reconstruction and medical imaging applications are executed. Dedicated software has been developed to implement multiview capability. A number of static or animated contemporary views of the same object can simultaneously be seen on the 3D-LCD screen by several observers, having a real 3D perception of the visualized scene without the use of extra media as dedicated glasses or head-mounted displays. Developed software applications allow real-time interaction with visualized 3D models; didactical animations and movies have been realized as well.

Light Field Visualization for Training and Education: A Review

“…Early works proposed holography in augmenting medical and anatomical education [153,154]. Several attempts for multiview autostereoscopic display utilization in medicine were carried out, including the implementation of a software for interacting with 3D medical models in real time [155], the 3D visualization of real captured surgical videos [156], neurosurgical reviews [157], an educational application for head and neck anatomy [158], and interactive 3D torso anatomy [159]. In a recent attempt to use autostereoscopic displays, specifically the 8-inch Looking Glass Portrait display [160], King's College London (KCL) implemented an online platform, namely Virtual Anatomy and Histology (VAH), targeted for anatomical teaching by means of 3D visualization of medical models and medical scans [161].…”

SPIE Medical Imaging 50th anniversary: historical review of the Image-Guided Procedures, Robotic Interventions, and Modeling conference

“…Throughout the formative years of the conference, advanced image visualization remained an important theme. The development of technologies and techniques to enable multi-modal image manipulation, visualization, and display led to the advent of virtual, augmented and mixed reality applications in medical imaging, with several notable examples being holographic stereograms, 26 real-time auto-stereoscopic visualization, 27 use of stereo and kinetic depth cues for augmented reality of brain imaging, 28 as well as the use of solid models of patient specific anatomy generated from computed tomography/magnetic resonance imaging (CT/MRI) images using laser sintering and laminated object manufacturing techniques. 29 …”

Multi-view point splatting

“…such as [SGI ], [Portoni et al 2000], [Schmidt and Grasnick 2002] and [Miniel et al 2004] is to render the 3D data in N passes using N different perspective (off-axis asymmetric sheared) view frustum configurations according to the multi-view setup as indicated in Figure 5. The resulting N images must then be combined into a spatial image conforming to the auto-stereoscopic display device.…”

“…In stereo rendering environments, generally the displayed 3D data set is rendered twice, once for each left/right eye view. Consequently, as done to date the N views need to be rendered individually, and combined and masked for a multi-view auto-stereoscopic display [Portoni et al 2000].…”

“…An extremely high computation and graphics power is required to do so with conventional approaches. Without the exploitation of similarities, rendering two views (for stereo) can require already twice the rendering time t of a mono-view rendering algorithm [He and Kaufman 1996;Ebert et al 1996;Koo et al 1999;Portoni et al 2000;Miniel et al 2004;Wan et al 2004]. Correspondingly, an N -view display requires a rendering time of up to N · t. Even exploiting modern graphics hardware (GPU) acceleration, this overhead prohibits effective multi-view rendering of large geometric data sets.…”