Automated Classification of Fetal Cardiotocograms for Antepartum Monitoring

The classification of fetalcardiotocograms (CTGs) is critical for antepartum monitoring, aiming to ensure fetal well-being and timely intervention. Current methods face challenges such as imbalanced data distribution and limited classification accuracy, necessitating robust solutions for reliable clinical application.This research introduces the Fetal Cardiotocograms Classification Network (FCC-Net), a novel framework designed to address these challenges. The methodology begins with extensive preprocessing of CTG data to remove noise and standardize features, ensuring optimal input quality. The Synthetic Minority Oversampling Technique (SMOTE) is employed to enhance class distribution to counter the issue of imbalanced data, enabling effective learning. For classification, the Light Gradient Boosting Machine (LGBM) is adopted, leveraging its high computational efficiency and accuracy for multiclass problems. The proposed FCC-Net demonstrates exceptional performance, achieving an accuracy of 99.1525%, precision of 99.8343%, recall of 99.7633%, and F-score of 99.3639%. These results highlight the potential of FCC-Net to significantly improve the reliability of CTG classifications, making it a valuable tool for clinical decision-making.