Fuzzy membership estimation using ANN: A case study in CTG analysis

Abstract

The Cardiotocograph (CTG) is being used by the obstetricians since 1960s as a means for recording (graphy) the heart beat (cardio) and the uterine contraction pressure (toco) of the mother, to evaluate the well being of the fetus. One of the major features of fetal heart rate (FHR) is its baseline,the accurate classification which is of utmost importance as all the other parameters of CTG rely on it. Inherent vagueness in the assessment given by the physicians can probably be modeled using fuzzy logic. It is one of the most trusted tools to handle uncertainty intrinsically present in the linguistic expression of human. The main challenge in designing a fuzzy logic based system is to design its membership function. In this paper we have presented a ANN based technique for the design of Fuzzy Membership Function (FMF) of FHR and used it in Fuzzy Unordered Rule Induction Algorithm (FURIA) in order to classify the CTG. The results obtained show significant improvement in classification over non FMF based technique.