Fetal heart rate analysis using a non-linear baseline and variability estimation method

Automated analysis of the fetal heart rate (FHR) curve plays a significant role in computer-aided fetal monitoring. Our study proposed a novel analyzing system of FHR signals which is constituted of FHR baseline estimation module, FHR acceleration/deceleration detection module, and FHR variability estimation module. In baseline estimation module, a novel non-linear FHR baseline estimation method using empirical mode decomposition and computational intelligence method of Kohonen neural network (KNN). We also designed a time-domain detection method of accelerations and decelerations according to international standards. Then the FHR variability estimation module was also designed using a combination method of empirical mode decomposition and moving-average filtering methods. We designed quantitative methods to test the performances of baseline and variability estimation. The results show that the new analysis system reaches a mean subjective satisfaction rate of 96.5% above medium, in terms of FHR baselines. The basal FHR values are close to the estimations from the experts, with a mean absolute error of 2.19 bpm. The acceleration/deceleration detection rate rises with the new baseline estimation method adopted. Besides, the long-term variability estimations are also close to those of the experts' with an MAE of amplitudes of 1.77 bpm and an MAE of cycles of 0.51 cpm.