Research And FPGA Verification Of HHT Algorithm Applied To Fetal Heart Sound Processing

Monitoring the health status of the fetus,giving timely correction and treatment before the occurrence of miscarriage or other dangerous conditions,realizing the healthy pregnancy and smooth production of pregnant women,is the significance of fetal health supervision.At present,fetal monitoring mainly uses ultrasound and ultrasonic Doppler technology in hospitals,this method has many disadvantages.Because of the economic development and the differences between urban and rural areas,the demand for pregnant women can't be met,and a great variety of examinations have also led to hospital personnel congestion and increased maternal family burden.In this paper,a method of extracting fetal heart sounds based on passive signal acquisition is proposed.The collection method is natural and safe,suitable for community and family type fetal health monitoring,and the result of fetal heart sounds is clear and intuitive.It has important social value and significance.Firstly,according to the characteristics of the fetal heart sound signal in time-frequency domain,several different processing methods are compared.On the basis of optimizing Hilbert Huang transform(HHT)and combining the position distribution law of fetal heart sound in heart sound cycle,a method of fetal heart sound extraction is designed.The method is used to extract and remove noisy fetal heart sounds on MATLAB.Secondly,after considering the speed,hardware complexity and resource consumption of the heart sound processing method,the implementation of the logic circuit for each module of the algorithm is introduced.Each module's RTL code were completed by using Verilog HDL language.Finally,On the XILINX development platform NEXYS4 DDR(chip XC7A100T-1CSG324C),an online logic analyzer(ILA)is used to verify the results.The FPGA application in the verification platform is in the field of portable and low power consumption which requires high real-time.The actual results show that the processing time is lower than the signal sampling time and consumes about 1/3 of hardware resources.It is proved by experiments that the improved algorithm is used to deal with the fetal heart sounds.Results the precision loss is 10%~15%,the signal-to-noise ratio is improved by 38.7dB,and the signal characteristic of the processing result is clear.