Classification of doppler signals with artificial neural networks using wavelet transform and fractal dimension

Recently, functional corruptions in circulation system, embolism, venoconstriction and vasolidation have been coincided very often. As a result of these problems, severe ilnesses have occured and bloodstream problems can cause to death. To prevent these illness of fatal cases, The Ultrasonic Doppler Technique, based on analysing the body area where is considered to be diseased, is used without any surgical operation to patient.In this study, a new practical approach, based on feature extraction to classify Doppler Heart Signs, has been suggested. In this approach, wavelet transform and fractal dimension calculating is used in feauture extraction process and backpropagation artificial neural network is used in classification process. At the end of study, 85% accurate classification success has been obtained. Obtained results have been compared with the results of entropy calculation refers to reference [7]. Accordingly, it is proved that entropy calculation technique produces more reliable results than fractal dimension calculating technique.Key words: Biomedical Signal Processing, Doppler Ultrasound, Wavelet Transform, Fractal Dimension, Artificial Neural Network.