Fast cross-correlation method for real time detection of fetal heart rate

To detect fetal heart rate (FHR) with high accuracy and low computational complexity, we propose a method based on the cross-correlation technique. The method consists of two procedures. First, the cross-correlation between two adjacent segments of an input sample at various lengths is computed to obtain the pitch period. Second, the FHR is estimated by finding the location of the first maximum in the cross-correlation values. These procedures differ from the conventional methods in that the FHR is computed with an input sample shifted by a certain amount compared to the previous fetal heart period interval. In experiments, results from conventional methods, the autocorrelation function (ACF) method and the average magnitude difference function (AMDF) method are obtained and compared with the proposed method for the purpose of evaluating its performance, and we confirmed that the proposed method gives an improved FHR with high accuracy