3D Medical Illustration Approach using real Patient Dataset

Abstract

Traditionally, there are three primary 3D rendering techniques applied to medical data which are surface rendering, volume rendering and maximum intensity projection (MIP). However, current medical society looks into non-photorealistic rendering (NPR), which offers a more interesting way to represent 3D medical data. The NPR techniques can adapt hand drawn medical illustration style such as water colour and pen and ink illustration. Another advantage of medical illustration in NPR is the capability to effectively present information and provide a familiar environment to medical practitioner had been trained with similar images for years. However, most of the NPR methods purpose such pen and ink, hatching and stippling and cartooning are based on standard medical data set. This paper investigation to see the impact of NPR techniques representing a real patient’s CT scan dataset. New medical illustration shading is introduced, which adds an opacity variable to the traditional Phong shading. These illustration styles are implemented using the GL shading language, where each of the illustration algorithms is transformed to programmable vertex and fragment shader. Based on the questionnaire conducted, the medical illustration styles is preferred compared the traditional shading style. The overall rendering performances for all the medical illustration styles are above 37 fps which fulfill the real time rendering requirement.