CAD for Detection of Fetal Electrocardiogram by using Wavelets and Neuro-Fuzzy Systems

摘要

The instrument which is used to measure instantaneous foetal heart rate (FHR) and labour activity is known as cardiotocograph. The normal oscillation FHR shows healthy foetus and absence of FHR oscillation is sign of foetal distress. Various methods are used for cardiotocographic, such as invasive and non-invasive. The recording of electrical activity of heart is done by electrocardiograph (ECG). The proper analysis of foetus ECG is important to detect the foetus condition inside abdomen to save foetus life as earlier as earlier possible. Here we use non-invasive method i. e abdominal foetal electrocardiogram (AFECG) to analyse the foetus internal condition. The invasive is harmful to mother and foetus both and need lot of precautions. Due to use of various instruments and improper handling, there is chance of addition of noises. Computer added diagnosis (CAD) is used, since in this there is no need of expert radiologist and chance of human error is zero. Noises can be removed by proper techniques to improve signal to noise ratio (SNR). The foetus heart starts to work after 21 days of pregnancy. So early detection is possible with high SNR. The maximum amplitude of foetus ECG (FECG) recorded during pregnancy is 100 to 300 μV. Which is vary less in compare to mother ECG (MECG), which is about 1 mV. Here we use wavelet transforms and artificial intelligence systems to de-noise composite signal and to obtain FCEG from AFECG. All coding is done in MATLAB software. First similar signal as obtained from mother abdomen and thoracic by leads are generated by coding. The obtained signal is trained by membership function before and after. The artificial neural network and fuzzy interference system (ANFIS) is used to obtain exact FECG. This can be implemented in real time system. The overlapping QRS parts are segmented finally by wavelet transform. The obtained result at various stage is sufficient to distinguish FECG and MECG from obtained signal. The effect of noise is also minimised and better signal to noise ratio is obtained. It is very useful for early diagnosis and to save foetus life. The observation during labour and delivery is done by monitoring FHR. It is also useful in biomedical telemetry and telemedicine.