Insights Into Three-Dimensional Ultrasound Mosaicing Strategies for the alignment of multiple ultrasound volumes

The creation of 2D ultrasound mosaics is becoming a §common clinical practice with a high clinical value. §The next step coming along with the increasing §availability of 2D array transducers is the creation §of 3D mosaics. In the literature of ultrasound §registration, the alignment of multiple images has §not yet been addressed. Therefore, we propose §registration strategies, which are able to cope with §problems arising by multiple image alignment. We use §pair-wise registration with a consecutive Lie §normalization and simultaneous registration, which §urges the usage of multivariate similarity measures. §In this thesis, we propose alternative multivariate §extensions based on a maximum likelihood framework. §Due to the higher computational cost of simultaneous §registration, we describe possibilities for speeding §them up, among others, the usage of a stochastic §pyramid.§§We also present methods to reduce speckle noise and §detect shadow in US volumes. For the compounding of §the final volume, a variety of approaches are §listed. Experimental results show the good §performance of the proposed registration strategies §and similarity measures.