Binarized spiking neural network optimized with momentum search algorithm for fetal arrhythmia detection and classification from ECG signals

Abstract

Diagnosing the fetal cardiac abnormalities by fetal electrocardiogram (FECG) is a difficult task, but it is essential to identify the fetus health condition. FECG monitoring is essential to provide accurate information on the state of the fetus. Severe fetal arrhythmia can cause heart failure or death. Therefore, it must be required to identify fetal arrhythmias early. The obtained ECG signals are very noisy and have artifacts from breathing and muscle contraction, which makes ECG extraction difficult. Misdiagnosis of fetal arrhythmias can lead to inappropriate treatment, which may result in further complications. If a defect has occurred in the heart of the fetus, it leads to death. To overcome these issues, Binarized spiking neural network optimized with Momentum search algorithm for Fetal Arrhythmia Detection and Classification from ECG signals is proposed in this manuscript for classifying the output as normal and arrhythmia. Initially, the input ECG signals are considered as Fetal ECG data base. The pre-processed ECG signals are extracted by Hexadecimal Local Adaptive Binary Pattern (HLABP). These extracting features are given to BSNN classifier for categorizing the output as normal and arrhythmia. Generally, BSNN does not adopt any optimization methods to determine the optimal parameters and to guarantee an exact classification. Hence, Momentum search algorithm (MSA) is employed to enhance the weight parameters of BSNN. The proposed technique is activated in python and utilizing some performance metrics, like sensitivity, precision, recall, f-measure, specificity, accuracy compared with existing approaches.