Three-dimensional spatial compounding of ultrasound images.

One of the most promising applications of 3-D ultrasound lies in the visualization and volume estimation of internal 3-D structures. Unfortunately, the quality of the ultrasound data can be severely degraded by artefacts and speckle, making automatic analysis of the 3-D data sets very difficult. In this paper we investigate the use of 3-D spatial compounding to reduce speckle. We develop a new statistical theory to predict the improvement in signal-to-noise ratio with increased levels of compounding, and verify the predictions empirically. We also investigate how registration errors can affect automatic volume estimation of structures within the compounded 3-D data set. Having established the need to correct these errors, we present a novel reconstruction algorithm which uses landmarks to register each B-scan accurately as it is inserted into the voxel array. In a series of in vitro and in vivo trials, we demonstrate that 3-D spatial compounding is very effective for improving the signal-to-noise ratio, but correction of registration errors is essential.