AN INTELLIGENT PATTERN RECOGNITION SYSTEM BASED ON NEURAL NETWORKS AND WAVELET DECOMPOSITION FOR INTERPRETATION OF LUNG SOUNDS

Ibrahim TURKOGLU ¹

In this study, a new automated pattern recognition system for interpretation of lung sound signals based on wavelet decomposition of signals and classification using neural network is developed. Inputs to the system are the lung sound signals acquired by a stethoscope in a noiseless environment. Features for the objective concise representation of lung sound signals are generated by means of wavelet decomposition. Classification of the features is performed using a back propagation neural network with adaptive learning rate. Ten different types of lung sound signals are analysed. Fifty-one record windows obtained from young humans are studied. Thirty of the record windows in database are selected to be used for training phase of the neural network. The results show that the intelligent pattern recognition system used to interpret ten different types of lung sound are acquired a high success.

Keywords
Lung Sounds, Wavelet Decomposition, Neural Network,