Fetal Heart Rate Classification: First vs. Second Stage of Labor

Fetal Heart Rate (FHR) is clinically used for early detection of fetal acidosis. Despite a marked interest in automatic detection procedures, FHR analysis remains a challenging signal processing task, receiving considerable research attention. Among other difficulties, the two stages of labor (dilation and active pushing) produce very different FHR dynamics. Most research efforts, however, have either ignored these differences or analyzed only one of the two stages of labor. In this work, we propose to assess the impact of labor stages on acidosis detection performance. A state-of-the-art sparse support vector machine classifier that performs simultaneously feature selection and classification is applied to a large-size and well documented FHR database. It shows that the selected set of features differs for each stage and that detection performance improves when the difference between labor stages is considered.