Shape Pattern Based Sinusoidal Fetal Heart Rate Detection from Scanned CTG Records

Fetal heart rate (FHR) monitoring is a widely used method for fetal health assessment. At present, most of the FHR data are recorded on the cardiotocogram (CTG) paper. Based on the morphological shape of the FHR in the CTG image, pathological patterns are diagnosed by obstetricians with the experience. However, this method lacks a unified evaluation standard. Consequently, it is necessary to construct a computer-assisted diagnosis method. Thus, in this paper, considering that the sinusoidal FHR (SFHR) pattern indicates severe conditions such as fetal hypoxia, the SFHR pattern is modeled in CTG image format based on the characteristics of periodicity and fractal dimension. Meanwhile, an improved linear SVM with parameters optimization is designed to ensure the full recognition of the SFHR. In addition, the linear model is further optimized by the fluctuation limit to reduce the misrecognition. Finally, based on the real FHR data source, the simulation shows that the model has a good performance and the classification accuracy is more than 97%.