DETECTING CHOLESTEROL USING NEURAL NETWORK AND FUZZY LOGIC IN ULTRASOUND IMAGING

Death caused by stroke above the age of 60 years is placed second in the world, and is the fifth leading cause in the people aged 15 to 59 years old. The formation of cholesterol in carotid artery is the main cause of stroke. A non invasive method of cholesterol identification in ultrasound scanning has been implemented in this paper. The cholesterol in the carotid artery is identified by measuring the thickness of the intima and media layer in it. Major focus of this proposed work is to build software to trace edges automatically and identifies the intima media thickness of the carotid artery using 3rd order polynomial equation and to classify the abnormality in the images using neural network. Fuzzy principles are embedded to overcome uncertainty parameters such as pixel size variation, colour variation and orientation problems in the image processing. By this project the machine automatically detects the desired output and does not finalise the scan until the correct image is acquired. Thus faulty calibrations by human errors can be avoided and also by embedding this feature in an ultrasound machine will eliminates the necessity of trained sonographer in operating the machine. Thus this non-invasive sonographic examination of carotid artery has its potential in clinical practice in early detecting skin cholesterol and cardiovascular diseases.