The volume rendering technique for 3D display of organ surfaces from 3D data array was found to be superior to the other techniques in image quality. Surface-shading calculations are performed at every voxel with local gradient vectors serving as surface normals. In the separate step, nonbinary surface classification operators are applied to compute a partial opacity for every voxel. The resulting shading and opacities are then composited along viewing rays to form an image.
In this thesis, we have implemented 3D surface display from ultrasound volume data. Ultrasound data are so noisy that we cannot use gradient intensity to enhance surface voxels. We defined the surface of ultrasound as a region with height and width greater than some value. We selected the peak of gray value as the most probable surface and adaptively applied variable opacity function to each pixel with its gray value.
The performance of the proposed volume rendering algorithm is investigated by computer simulation using ultrasound volume data collected as two-dimensional planar images with commercially available equipment. In contrast to the conventional volume rendering methods, the proposed method reduced artifacts and blurring effects of resulting image.
