The complex and interrelated trunk asymmetries due to scoliosis are difficult to quantify succinctly. Changes in topography can be quantified independent of both orientation and growth by using the curvatures inherent in the trunk surface. Digital representations of the trunk surface were produced by projecting a pattern on the otherwise featureless trunk. This paper reviews approaches to representing surface shape. An approximation technique was used to quantify the curvature from the discretely sampled surface. Curvature is a combination of both the first and second order partial derivatives. These derivatives were determined through convolution with a Gaussian kernel. Tradeoffs were necessary between kernel size and width of the Gaussian. The goal was to have the Gaussian large enough so a sufficient local area could be used to compute the partial derivatives while ensuring that the majority of the volume of the Gaussian was within the kernel. The surface was segmented into local regions of similar characteristics. A shape index was used to distinguish nine regions - spherical cup, trough, rut, saddle rut, saddle, saddle ridge, ridge, dome and spherical cap. We concluded that this shape index provides the most discriminating representation of trunk shape.