Computer Science (Jan 1999)
Methods of the Straightening Transformation and Analysis of Outer Contours of Objects in Medical Image Recognition
Abstract
The article presents an algorithm of the transformation straightening outer contours of objects, allowing to transform biological structures of interest, visible in medical images, from two-dimensional space of the image into two-dimensional graph, representing contours of the straightened object with its possible outer morphological changes, indicating the progress of pathological processes. The developed method is based on a sequence of geometric transformations of the analyzed objects in the image, and its operation was described in detail and illustrated by examples of transfonnation of main pancreatic ducts, appearing in X-ray images acquired by ERCP, as well as by graphs showing outer morphology of blood vessels in the fundus of the eye. The proposed method is universal, and due to the fact that the transformation preserves and exposes pathologic morphological changes, such as abnormal lateral branches, local dilatations or projections it can also be used in the recognition and analysis of other structure images acquired by various techniques of computer tomography, e.g. cerebral vessels, bile ducts, coronary vessels or fragments of gastrointestinal tract. The graphs of the structures of interest obtained as a result of the transformation straightening outer contours of objects are a starting point for the recognition of morphological changes, important for diagnosis, using syntactic methods of pattern recognition.
