Unsupervised Clustering for Fetal State Assessment Based on Selected Features of the Cardiotocographic Signals

In modern obstetrics the cardiotocography is a routine method of fetal condition assessment based mainly on analysis of the fetal heart rate signals. The correct interpretation of recorded traces from a bedside monitor is very difficult even for experienced clinicians. Therefore, computerized fetal monitoring systems are used to yield the quantitative description of the signal. However, the effective techniques enabling automated conclusion generation based on cardiotocograms are still being searched. The paper presents an attempt to diagnose the fetal state basing on seventeen features describing the cardiotocographic records. The proposed method applies the unsupervised classification of signals. During our research we tried to classify the fetal state using the fuzzy c-means (FCM) clustering. We also tested how the efficiency of classification could be influenced by application of principal component analysis (PCA) algorithm. The obtained results showed that unsupervised classification cannot be considered as a support to fetal state assessment.