Analysis of tomographic volume imagery would be greatly facilitated if the objects within the volume could be presented in three-dimensionally (3D) rendered views. Such a capability has not been developed in large part because the problem of image segmentation remains unsolved. We describe a new approach that circumvents this problem by allowing the human user to segment images interactively using morphology functions. This segmentation is performed concurrently with 3D visualization providing direct visual feedback to guide the user in the segmentation process. Thus, rather than attempting to duplicate the complex and poorly understood human pattern recognition capability, our approach uses the human's own judgment and knowledge. We present a research study to demonstrate the general feasibility of this approach using MR and CT images. The massive data and computational requirements for interactive 3D image processing exceed current processor limits, but the increased capacities of the next generation of computers are expected to make this approach practical.