Three Dimensional Image Presentation Techniques in Medical Imaging

Medical images can be presented three-dimensionally by techniques that either calculate the effect of reflections from surfaces predefined from slices or project a three-space of luminosities computed from voxel intensities onto the visual receptors. Slice-based reflective displays are the most common type. Means of producing surface descriptions both via voxel sets and via slice contours will be reviewed. Advantages of and means of transparent disylay to allow the appreciation of the 3D relationships among objects wil be set forth. Ways to prOduce additional depth cues by stereoscopy and the kinetic depth effect will be discussed, and the importance of interactive modification of viewpoint, clipping plane, displayed objects, etc. will be explained. A new device, UNCs Pixel-planes, for accomplishing this in real time will be illustrated. Voxel intensity based display methods avoid the need for time-consuming predefinition of object surfaces and thus can allow exploration of 3D image data. V arifocal mirror hardware and fast computation of one or more projections based on object probabilities are two of the more im~rtant a_pproaches. While 3D display provides important mformabon about 3D relationships, it cannot provide the kiild of appreciation of subtle grey-scale changes that 2D display can. Methods that can combine these two kinds of information by superimposing 20 grey-scale slices on or in the context of 3D displays will be discussed. APJ?lications of these techniques for both diagnosis and radiotherapy planning Will be used as illustrations and guides to the usefulness of these techniques with CT, MRI, and other 3D medical imaging modalities. Introduction Modern medical imaging modalities, such as CT and MRI, produce image data in slices, and these are conventionally viewed as an array of 2D grey-scale slice images. The advantages 'of comprehension in three dimensions have spurred the development of methods of 3D display of this data. 3D display of single, high contrast objects, such as bones imaged by CT, is beginning to be widely used in both diagnosis and surgical planning. Moreover, early systems for the display of lower contrast objects, such as soft tissue imaged by CT, and of the relationships of multiple objects, both anatomic and computed, show considerable promise for radiotherapy treatment planning, surgical planning, and diagnosis. This paper will survey approaches to effective 3D display to achieve these ends. 3D display can be roughly divided into two approaches. In the frrst, presently more common, approach a surface is determined from the original recorded image slices, and reflections of light impinging on this surface are simulated. We will call this approach slice-based reflective display. In the second approach the display is produced directly from the original recorded intensities, thought of as volume elements (voxels) . Some combination of transparency, reflectivity, and (possibly colored) luminosity is attributed to each voxel, and the transmission through and reflection by these voxels from a chosen set of light sources is simulated. We will call this approach voxel intensity based display. We will focus on methods of slice-based reflective display because of their much more • common use, but we will follow their description with a short summary of voxel intensity based methodology. We will also spend some attention on visualization methods that are applicable to both types of display.