Bone Crack Detector based on X-Ray using Fuzzy Logic and Neural Network
No Thumbnail Available
Date
2013
Journal Title
Journal ISSN
Volume Title
Publisher
Uva Wellassa University of Sri Lanka
Abstract
An x-ray (radiograph) is a noninvasive medical test (Tian Tai Pengn, 2002) that helps
physicians diagnose and treat medical conditions. X-Ray images are used by doctors to detect
the crack and abnormal conditions of the bones. Doctors are analyzing thousands of X-Ray
images at hospital day by day. That activity is monotonous and also consuming lot of time.
Bones contain much calcium, which due to its relatively high atomic number absorbs X-Rays
efficiently. This reduces the amount ofX-rays reaching the detector in the shadow of the bones,
making them clearly visible on the radiograph.
A recognition system has three parts; Image Processing and feature extraction, Fuzzy Logic
based identification and Neural Network based verification. The main objective of this study is
Computer-assisted decision-making system to detect the crack of the bone in X-Ray image.
Image Preprocessing is applying for enhance the features of the image. Edges based filters apply
for enhance the edges because edges perform the vital role for detecting the crack of the bone
image. Then system detects the edges of the image using canny edge detector.
Background of the image is eliminated as the next step. System is finally detecting the edges
that could be a crack or not. System could detect the actual crack and also some specific
features of the bone. Those specific features of the bone are smooth lines and cracks are rough
line. Using that specific characteristics system separate crack lines form some features of the
bone. Finally abstract the features information for Fuzzyfication.
Fuzzy Classifier contains fuzzy inference engine, input output variables. Input variables are
information about the edges. Output variables are the detected crack. System is using two fuzzy
sets and three fuzzy functions for each fuzzy set. One fuzzy set is Fuzzy Multiplication and
other is Fuzzy Ratio. The fuzzy rules calculate the output and those outputs send to Neural
Network for verification.
Neural Network (Davis et al., 1999) takes the input from the Fuzzyfication and specific some
parameters taken from image. Eight neurons for input layers, ten neurons for hidden layer and
three neurons for output layer use for Neural Network. Supervise training uses for trains the
Neural Network. Output obtains as pattern. Finally Neural Network verifies the fuzzy output
and correctly says crack or not a crack.
Description
Keywords
Science and Technology, Computer Science, Networks, Technology, Health Science