Coarse segmentation of cervical and lumbar vertebrae using a customized version of the Generalized Hough Transform
Arunkumar Gururajan · ThinkTech (Texas Tech University) · 2003
The objective of this thesis is to develop a framework for a robust coarse segmentation scheme, to segment the cervical and lumbar vertebrae in radiographic images. Previous research customized the Generalized Hough transform (GHT) for segmenting the radiographic images of cervical vertebrae, and tested the same on a dataset of 50 images. The key factor for the successful performance of GHT, is the availability of suitable edge information for the target shape that is to be segmented. This thesis is an effort to effectively address the low-level vision schemes involved in the preprocessing stage of GHT, and also to extend and test the proposed coarse segmentation scheme on a sizeable set of cervical and lumbar spine images.