Thresholding reflux episodes in impedance measurements using a neuro-fuzzy system

Multiple intraluminal impedance measurements (MII) provides impedance signals which reflect the changes in the electrical conductivity inside the gastrointestinal tract (GIT). To determine a reflux episode from MII signals, wavelet transform is used to detect the edges of the impedance, then thresholding impedance at certain level. A predictor for the value of the threshold is constructed by training a fuzzy system using a number of features. During training process the predictor clusters the features into classes of similar feature values. The trained system is used to predict the threshold applied on the distal channel. Subsequently, a combination of wavelet transform and fuzzy logic is used to classify the episodes. In this paper we present the results of one stage of the system that predicts the threshold applied on the impedance value.