Skip to Main Content
This paper presents a framework for the rapid isolation and representation of structures from spatially registered image stacks. The introduced framework enables dynamic visualization with next generation volumetric displays, in which image volumes are rendered in real 3D space. Two recent semi-supervised approaches to the problem of segmenting relevant structures are investigated and integrated to yield a generic interactive visualization pipeline. Information obtained from the segmentation process is used to guide layered object reconstruction via the SpatialGL SDK. The reconstructed original objects in 3D may then be selectively visualized on the Perspecta: a spatially-rendered display. Practical considerations, future improvements and applications are discussed.