Automated Extraction of Fetal ECG Signal Features Using Twinned Filter and Integrated Methodologies

Nowadays, detecting fetal electrocardiogram (FECG) signals from a mother’s abdominal region is considered a most challenging task because of the high maternal overlapping and fetal signals at this location. Different existing FECG signal extraction techniques have been emphasized for extracting fetal ECG signal features. However, the noises present in the fetal signals still affect the overall performance; also, there is a high chance of missing waveforms and degraded signal-to-noise ratio (SNR) performance. This work proposes a novel extraction and detection technique to overcome these issues. At the initial stage, the FECG signals are collected from two datasets the abdominal and direct fetal ECG database and the Noninvasive fetal ECG database. The signals in the raw dataset with high noises are pre-processed using the twinned Savitzky-Golay filtering (Twin_SGF) model that can effectively enhance the SNR performance. After pre-processing, the FECG signals are fed into the feature extraction technique named Q-Integrated rapid-tune wavelet transform with independent component analysis (Q-IWavCA) to extract essential features from FECG signals. In addition, key characteristic information can be analyzed using the Extreme wavelet genre method (X_WavG) technique. This technique can detect P, T, QRS and ST segments effectively. The performance of a proposed method is analyzed via the MATLAB platform. The effectiveness of the proposed approach is proved by comparing it with different existing approaches in terms of sensitivity (99.3 and 99.5%), positive predictive value (PPV) (99.6 and 99.4%), F-measure (99.2 and 99.54%) are obtained for ADFECG dataset and NIFECG dataset correspondingly.