At the present, most medical volume rendering implementations are based on slice-based methods where axisor
Viewport-aligned textured slices are blended together to approximate the volume rendering integral. However, slicebased implementations are rasterization-limited and difficult to optimize. Moreover, when applying such transfer functions as perspective projection, the integration step size will vary along viewing rays when using planar proxy geometries, leading to visible artifacts. As a result, slice-based techniques cannot present an optimal volume visualization framework [1].