Surface and volume rendering in three-dimensional imaging: a comparison

Many surface rendering techniques are currently available for the three-dimensional display of structure data captured by imaging devices. Comparatively fewer volume rendering techniques are also available for the same purpose. The relative performance of these two methodologies in visualization tasks has been a subject of much discussion recently. Although it is very desirable to establish, based on observer studies, objective guidelines stating the relative merits of the two methodologies even for specific situations, it is impossible to conduct meaningful observer studies that take into account the numerousness of the techniques in each methodology, and within each technique, the numerousness of the parameters and their values that control the outcome of the technique. Our aim in this article is to compare the two methodologies purely on a technical basis in an attempt to understand their common weaknesses and disparate strengths. The purpose of this article is twofold--to report a new surface rendering technique and to compare it with two volume rendering techniques reported recently in the literature. The bases of comparison are: ability to portray thin bones; clarity of portrayal of sutures, fractures, fine textures, and gyrations; smoothness of natural ridges and silhouettes; and computational time and storage requirements. We analyze the underlying algorithms to study how they behave under each of these comparative criteria. Our conclusion is that, at the current state of development, the surface method has a slight edge over the volume methods for portrayal of information of the type described above and a significant advantage considering time and storage requirements, for implementations in identical environments.