Neural Network for Detecting Anterior Osteophytes

Abstract: Development of a computer vision system to identify degenerative disorder of the spine is a challenging task. An x ray image shows the features associated with these disorders. Manual analysis from observation of the x rays can be prone to errors due to the low contrast of the x rays images as well as variability in visual judgment from person to person. In this paper a neural network based approach for detection of osteophytes is investigated. Convex hull features of the segmented vertebrae are determined by image processing technique. The network is trained by conjugate gradient descent back propagation algorithm and tested on a set of vertebrae images. After the initial manual assistance in formation of template and the training of the network, the system becomes automated for real time detection of anterior osteophytes.

Keywords: Anterior Osteophyte, x-ray, segmentation, Neural Network, back propagation.