Automatic registration of 3-D ultrasound images

One of the most promising applications of 3-D ultrasound (US) lies in the visualisation and volume estimation of internal 3-D structures. Unfortunately, artifacts and speckle make automatic analysis of the 3-D data sets difficult. In this study, we investigated the use of 3-D spatial compounding to improve data quality, and found that precise registration is the key. A correlation-based registration technique was applied to 3-D ultrasound data sets acquired from in vivo examinations of a human gall bladder. We found that the registration technique performed well, and visualisation and segmentation of the compounded data were clearly improved. We also demonstrated that an automatic volume estimate made from the compounded data (13.0 mt) was comparable to a labour-intensive manual estimate (12.5 mL). In comparison, automatic estimates of uncompounded data are less accurate (ranging from 13.5 mL to 16.7 mL). The registration technique also has applications in intra- and interpatient comparative studies. (C) 1998 World Federation for Ultrasound in Medicine & Biology.