FPGA Design And Implementation Of FECG Extraction Based On Independent Component Analysis Algorithms

Fetal electrocardiogram (FECG) includes lots of information on the health of the fetus, andcould help doctor make diagnose during gestation and labor. However, the signals containmaternal electrocardiogram (MECG) signal, maternal myoelectric signal, and power lineinterference, besides FECG signal. FECG signal is often overwhelmed in the noise signals. Itis still a hot spots that how to extract pure FECG signal. Independent components analysis(ICA) is a new technology developed in the1990s, which can separate the mixture of severalindependent signals. Thus, ICA could be used to extract the FECG signal. The experimentindicates that there are many advantages to extract the FECG signal using ICA.This paper designs and implements the ICA algorithm with fixed point format data onFPGA to extract the FECG signal. This paper implements the modules involved in the ICAalgorithm, including matrix multiplier, matrix inversion, matrix eigenvalue decomposition andso on. The frameworks of FastICA proposed in the paper utilize the technology of pipelineand parallel computing to improve the computing speed and to reduce the consumption ofhardware resource. Some features should be emphasized:1. Implementation of digital signals processing on FPGA could speed the computation withlow consumption of hardware resource.2. Implementing the ICA algorithm with code is more flexible and efficient than that withmodeling the algorithm on high level.3. Real-time processing. Implementing the ICA algorithm on FPGA has high computingspeed, which could not only enhance the effect of extraction, but also ensure the real-timeprocessing.