Medical image rendering

Three-dimensional (3-D) visualization has recently become an established discipline in medicine. Although numerous visualization methods are currently available, a unified framework to describe and study them has been lacking. Often, the functionally independent operations in a method are integrated among themselves or with the method itself for computational efficiency. The two main aims of this article are (1) to review the methods in a unified way in a general setting so that it becomes possible to appreciate the interrelationship and interdependence of methods, and (2) to show how a variety of new methods emerge with potentially improved renditions in this unified treatment. To this end, we introduce an operator notation to describe concisely the basic 3-D imaging transforms commonly used in visualization and identify a comprehensive set of basic transforms. We describe several new basic transforms for filtering and interpolating structures and scenes, and for rendering surfaces and volumes. We show the power of the principle of treating 3-D imaging methodologies as comprising an appropriate combination of the basic operators. We show how such a treatment leads to a great variety of new rendering methods and how many such methods can lead to improved portrayal. We develop separate transform sequences to optimally render robust and frail structures (ie, structures represented in scenes with well-defined and ill-defined boundaries, respectively).