Reconstruction of biliary structure in 2D MRCP images using multi-scale analysis

Magnetic resonance cholangio pancreatography (MRCP) has become a reference technique for biliary tree analysis. Typical MRCP images, however, suffer from difficulty in distinguishing the structure of the biliary tree in order to identify abnormalities, for clinical diagnosis. For efficiency in analysing MRCP image series, the need arises for the use of semi-automated image processing techniques. A segment-based multi-scale approach is described, incorporated with image selection, enhancement and watershed segmentation, to identify and reconstruct the hierarchical biliary tree structure in 2D MRCP images. The results achieved may be further extended to higher dimensional images.