Asymmetry Identification Using Generalised Symmetry Axes

Abstract

We preset an algorithm for identifying asymmetry in CT images of the human body, relying on the local reflective axis of symmetry. As asymmetry is such images may be a sign of abnormality, our method compares the two sides of the image divided by the extracted axis of symmetry in order to identify regions of interest. Both, extraction of local axis of symmetry and identificaiton of deformities, are obtained automatically. The algorithm is based on registration and extraction of bone contours to generate the desired symmetry axis. Also, to facilitate comparison of the left side with the right side of the image, the local axis of symmetry is made straight by alignment of the image pixels.