Hierarchical σ-octree for visualization of ultrasound datasets

There are two important factors to visualize ultrasound datasets using volume ray casting method. Firstly, efficient methods to skip over empty space are required. Secondly, adequate noise-detection methods are necessary because ultrasound datasets contain lots of speckle noises. In general, space-leaping and noise-filtering methods are exploited to solve the problems. However, it increases the preprocessing time to generate the filtered datasets, and interesting (meaningful) objects could be affected by a filtering operation. We propose a hierarchical sigma-ctree containing min-max values and standard deviation for each block, named a hierarchical sigma-octree. In rendering step, our method refers to min-max values of a block. If the block is regarded as nontransparent, it also checks its standard deviation value to detect speckle noises. Our method reduces rendering time compared with the method using only the min-max values because most blocks containing speckle noises are considered as transparent.