Prognostic model based on image-based time-frequency features and genetic algorithm for fetal hypoxia assessment.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Obstetrics

Background:

• Cardiotocography (CTG) is crucial for fetal monitoring, but interpretation variability and low specificity limit its effectiveness.
• Automated CTG analysis offers a promising solution to overcome current limitations in fetal well-being assessment.
• Accurate prediction of fetal hypoxia is essential to reduce neonatal mortality and morbidity.

Purpose of the Study:

• To develop and validate a novel prognostic model for predicting fetal hypoxia using cardiotocography (CTG) traces.
• To introduce an innovative image-based time-frequency (IBTF) analysis for enhanced feature extraction from fetal heart rate (FHR) signals.
• To evaluate the performance of the proposed model in improving classification accuracy for fetal hypoxia detection.

Main Methods:

• A novel image-based time-frequency (IBTF) analysis combining short-time Fourier transform (STFT) and gray-level co-occurrence matrix (GLCM) was developed.
• Spectrograms of FHR signals were generated using STFT, converted to grayscale, and analyzed for IBTF features (contrast, correlation, energy, homogeneity).
• A genetic algorithm was employed to optimize feature subsets for input into a least-squares support vector machine (LS-SVM) classifier.

Main Results:

• The proposed IBTF analysis demonstrated improved feature extraction for FHR signals compared to conventional methods.
• The prognostic model achieved a sensitivity of 63.45% and a specificity of 65.88% when combining conventional and IBTF features.
• Classification accuracy was enhanced by the inclusion of IBTF features in the predictive model.

Conclusions:

• The study presents a promising new approach for feature extraction in FHR signal analysis using IBTF.
• The developed prognostic model shows potential for more accurate prediction of fetal hypoxia from CTG data.
• IBTF features significantly contribute to improved classification accuracy in fetal monitoring.