RESULTS

Digital model accuracy

The 800,000 polygon digital model scanned at 100 µm was compared to the physical specimen by scaling on the computer screen and by using microscopic techniques respectively. The dimensions, geometry, and color were identical on the digital model to that of the specimen. Figure 6 is a composite image comparing the digital model with a photograph of the physical specimen. To the eye, even at increased magnification, it was not possible to differentiate the specimen and the digital copy. As accuracy of the methodology has been previously rigorously tested under demanding industrial conditions it was not deemed necessary to quantify the verification process. However, visual examination of two regions highlights examples of the accuracy of the digital model. Figure 7 illustrates the vidian canal; its structure can be seen easily in both the model and the image of the physical specimen. Figure 8 shows details of the abducens cranial nerve (VI) and the branch of the internal carotid artery (ICB). The only limitation of digital models such as shown is that they can only support a limiting amount of scaling (Fig. 9). The limitation of this scaling is related to two variables; the scanning resolution and polygon reduction (Beraldin et al. 1997).

Comparison of 50 µm and 100µm scan resolutions

Figure 10 was scanned at 50 µm resolution, and Figure 11 was scanned at 100 µm resolution. Both models support substantial scaling without facets appearing in the digital model. However, when the 100 µm digital model is scaled to more than 5x, facets become visible on high angle surfaces (Fig. 12). In contrast, the model scanned at a 50 µm resolution supports much greater scaling, to the point of exceeding the resolution of the texture map (Fig. 13).

Polygon reduction

As discussed earlier, to permit most researchers access to a version of this complex digital model that can be manipulated, it was necessary to reduce the number of polygons. Figure 3 is a digital model comprised of 50,000 polygons scanned at a 100 µm resolution. This represents a 98% reduction in the number of polygons from the initial 3 million polygon model and can only support a limited amount of scaling (1.5x) before details apparent in the physical specimen are either obscured or absent (Fig. 14).