Snapping a Cursor on Volume Data

In this paper we explore the architecture of GPU-based volume ray-casting to control the cursor movements on surfaces of volume data only on the basis of their sampled scalar values. This method does not rely on building a tight proxy geometry nor computing local iso-surface parametrization. In this setting only a few instructions should be included in a ray-casting shader: at every fragment the depth of the closest non-transparent voxel along each viewing ray is calculated and stored in the depth buffer. Its application in the manipulation of 3D medical images has, however, a broad spectrum. Two of them are presented. The proposed technique is very simple and fast, yet produces very nice and intuitive visual feedbacks.