PATTERN RECOGNITION OF DEFECTIVE BONES X-RAY USING ARTIFICIAL NEURAL NETWORK

Abstract

Recently, the applications of Pattern Recognition involved in many fields, starting from scientific research, medicine, reaching to the crime recovery by pattern recognition of finger imprints. For any bones doctor, bones X ray represents very useful and important part in the diagnoses level, sometimes it should be the key in surgical operations. This paper merges the two fields (pattern recognition and bones medicine) by training simple artificial neural network using back propagation algorithm to recognize five normal and five defective bones X ray images and then evaluates the recognition accuracy. From the training of network, it's clear that a higher number of hidden neurons increase the processing time. The shape of images plays an important role in recognition process. It's obvious that the network should choose the minimum number of hidden neurons such that it still determines the maximum performance of the network