A new concept for intraoperative matching of 3D ultrasound and CT

Matching of ultrasound images with CT or MRI scans is an awkward and unsatisfactory task when using conventional methods. Wide ranging differences in modality of ultrasound and CT/MRI require new techniques to be explored for successful alignment. Ultrasound images characteristically show comparable high noise ratio due to scattering inside the region of interest and the surrounding area. Additionally, shadowing and tissue dependent echo response time produce geometric artifacts. These image distortions are sophisticated to recover. Though image quality and geometric relationship are poor, ultrasound images show the potential for fast, low-cost, non-invasive and flexible image acquisition, predestinated for intraoperative application. The fusion of intraoperative ultrasound and preoperatively acquired CT/MRI images provides both, geometric invariance and flexible fast image acquisition, merging in a powerful tool for augmented three dimensional reality. In this paper we describe a completely new concept for alignment with abstaining from direct rigid or elastic matching of ultrasound to CT/MRI. Instead of placing those images in direct relationship, our approach involves a simulation of ultrasound wave behavior in order to predict B-mode images.