My Thesis (Abstract) - 2007 - TIF Usakti

Bone Degradation Disease Detection System Using Backpropagation Artificial Neural Network Method and Characteristic Representation Within Eigen Space

There are several methods for osteoporosis (bone degradation) detection, one of them is through osteoporosis images observation of X-ray or roentgen photos, which will then be analyzed manually by rheumatologists.The purpose of this thesis project is to create a system which could detect human osteoporosis by applying rheumatology principles, which state that the primary identifiable location is between the wrist to each and every tips of fingers.
This software’s work process involves three important processes, which are image processing, pixel reduction process, and artificial neural network process. First, an X-ray digital image (30 x 30 pixel) will converted from RGB color to grayscale, then thresholding and get the graylevel values. That values then will be normalizing to interval [0.1, 0.9], reducing in eigen space using PCA (Principal Component Analysis’s method), and then the result would be processing on backpropagation artificial neural network system as input to know what the prediction of disease from an X-ray had input before it. From testing with learning rate 0.7, and momentum 0.4, this system had had success rate up to 73 % - 100 % for non-learning data test, and 100 % for learning data test.

Keyword : osteoporosis, image processing, PCA, artificial neural network.

Download Full Paper :

Backpropagation and Eigen Space