Extracting Anatomical Structures, Primarily Vertebrae From Images Of The Human Spine

摘要

Medical professionals use Magnetic Resonance Images of the human spine for evaluating bone tumors, cysts and bulging or herniated discs in the spine. Often the curvature of the spine is used as a useful diagnosis measure. Features in Magnetic Resonance Images often have weak edges. In addition, images acquired through modern sensors may be contaminated by a variety of noise sources. Traditional noise removal techniques will also result in the removal of the weak edges. This paper presents an algorithm for extracting the vertebrae and inter-vertebral discs from the Magnetic Resonance Images of the lower back (lumbar spine). It also presents a method for calculating the curvature of the spine. The algorithms discussed in this paper have been implemented using a Java based Graphical User Interface. The interface titled ExACT - Enhancement and extraction of Anatomical ComponenTs provides the flexibility to work with several different image formats in addition to allowing the user to tailor the algorithm to a particular image.