OPTIMIZING FETAL HEALTH ASSESSMENT WITH AI-DRIVEN UMBILICAL CORD CLASSIFICATION VIA 2-D DOPPLER ULTRASOUND IMAGING

Abstract

The fetal umbilical cord is an essential organ that connects the fetal to the placenta, supplying oxygen and food while removing waste. It is constructed of blood vessels that are enclosed in a gelatinous material. Ultrasound imaging provides real-time images of inside organs using sound waves, which helps in prenatal diagnosis and fetal development monitoring. In this research, we aim to develop a novel Artificial intelligence (AI)-driven umbilical cord classification model through 2-D Doppler ultrasound imaging. We proposed an innovative Dwarf Mongoose tuned Versatile Random Forest (DM-VRF) for classifying the umbilical cord based on ultrasound images. Initially, we gathered a dataset with 2-D Doppler ultrasound images, to train our model. Image Normalization algorithm is implemented to pre-process the gathered data. We extracted the crucial features from the processed data using a histogram of oriented gradients (HOG). This innovative approach leverages the significant behaviors of the DM algorithm to refine the performance and the accuracy of a classification model. The suggested approach is implemented in Python software, during the result analysis phase, we employed various parameters such as accuracy, Receiver operating characteristic (ROC), precision and recall to evaluate classification effectiveness. To determine the efficacy of the proposed approach, we performed a comparison study with other existing methods. The experimental findings show that the suggested model performs better than traditional methods for using image analysis to categorize the umbilical cord.