Meeting Banner
Abstract #1842

Robust Feature Based Pre-Registration of 3D MR Image to 3D B-Mode Ultrasound Image of the Liver

Chi Jun Weon1, Woo Hyun Nam1, Jung-Bae Kim2, Youngkyoo Hwang2, Won-Chul Bang2, Jong Beom Ra1

1Electrical engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea; 2Samsung Advanced Institute of Technology, Yongin-si, Gyeonggi-do, Korea

The registration of MR images with real-time US images is beneficial in various clinical applications. In most of registration algorithm, a pre-registration scheme to obtain good initial transformation parameters is important. However, conventional methods usually require a time-consuming and inconvenient manual process for pre-alignment. Furthermore, if anatomical features are not clearly visible in an abnormal liver, it is difficult to perform the manual pre-registration procedure. In this paper, we propose an automatic and robust pre-registration algorithm of the liver. For reliable and robust registration, the algorithm uses patient-independent feature geometries such as IVC and liver surface rather than vessel bifurcations.

Abstract PDF

Keywords

abnormal accuracy accurate addition advanced align aligned aligning alignment allows always applications arrows automatic automatically automation avoid bifurcations bottom breath build carefully centerline centerlines clinical combining compensate component computation computed consistent consistently coordinate correspondences critical dataset datasets defined dependent described determination determine diagnosis dimension discrete displacement distance domain electrical error especially evaluate evaluation even exhalation existing expected experimental extract feature features fiducial final five geometric guided hold illustrations initial institute interval intervention king larger liver manual manually many maps mark matrix meanwhile measure mechanical medical mode mostly note observation optimization pairs paper patient performance perpendicular planning precise precisely principle procedure propose proposed proposes qualitative quantitative real reducing refer regardless register registered registration reliable respectively respiratory rigid robust robustly robustness rotating sample sampling science segment segmented selected semi several since started step steps straight subsequent surface surgeons surgical systems table target technology transform transformation transformed translated ultrasound uniform useful utilize vessel visible