Enhanced Cervical Dilation Assessment Using CNN ResNet 50 with Inception Module

Accurate and timely monitoring of cervical dilation is essential for evaluating labor progression in pregnant women. Traditional manual methods are often subjective, leading to variability in assessments and patient discomfort. While existing systems utilize low-light intensity imaging probes to improve visualization and reduce discomfort, manual interpretation of these images remains prone to inaccuracies. To address these challenges, the proposed system integrates deep learning with existing imaging techniques by employing a hybrid ResNet50 architecture enhanced with the Inception module. This approach enables effective classification of cervical images by leveraging residual connections for efficient feature extraction and multi-scale convolutions for capturing diverse patterns. The model was trained on an augmented dataset of 402 images derived from 134 original samples using techniques such as CLAHE, rotation, and brightness adjustment. Experimental results demonstrated high performance, achieving an accuracy of 96.4%, precision of 96.7%, recall of 96.4%, and an F1-score of 96.4%. By automating the evaluation process, the system minimizes subjectivity and enhances assessment accuracy, contributing to more reliable and efficient obstetric care.