University of Southampton
Automated segmentation of lumbar vertebrae for the measurement of spine kinematics
Abstract
dc:description.abstract<p>There have been a number of studies on vertebral extraction from fluoroscopic images, varying from manually locating the vertebral landmarks to template matching. However, the former method poses problems of repeatability which can lead to errors in kinematic analysis, while the latter cannot cope with the large changes in the illumination and contrast amongst the frames. In this thesis, extended forms of the Hough transform, together with a recent method of low level feature extraction (phase congruency), have been used to solve this problem. Phase congruency allows for better low level feature extraction, but even this is sufficient to reconcile errors which needs a higher level interpretation. The generalised Hough transform can be used to extract arbitrary shapes, but can suffer from discretisation errors. In the new approaches to the Hough transform, Fourier descriptors are used to describe the vertebral body shape. This description was incorporated within the Hough Transform algorithm from which the affine transform parameters, i.e. scale, rotation and center position can be obtained. The method has been applied to images of a calibration model and to images from a sequence of a moving human lumbar spine. The results are encouraging but sometimes difficulties can be experienced in the extraction because of the extremely poor image quality.</p> <p>A new spatio-temporal Hough transform has been developed which can improve performance by incorporating the contextual information within the image sequence. An energy function that represents the compromise between the contextual information and the Hough space has been constructed. Rather than finding the maxima in the Hough space like traditional Hough transform, here the Genetic Algorithm is employed to search for maxima of the energy function. The application results on nine subjects show better performance than the traditional Hough transform.</p>
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2003
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Zheng, Yalin
Rights
- Language dc:language
- English, eng