Artificial Intelligence Techniques for Semantic Segmentation and Classification for Ovarian Cancer Detection Using EfficientNetB0 with FPN and Causal Dilated Convolutional Neural Networks: Trends and Challenges

Abstract

Ovarian cancer remains one of the most critical gynecological malignancies, largely due to late-stage detection and the limitations of traditional diagnostic methods. Medical imaging techniques such as ultrasound, CT, and MRI are commonly used for tumor identification, but manual interpretation is often time-consuming and subject to variability among clinicians. In this context, artificial intelligence, particularly deep learning, has emerged as a powerful tool to enhance diagnostic accuracy and efficiency. Convolutional neural networks enable automated extraction of complex features from medical images, allowing models to detect subtle patterns that may not be easily visible to the human eye. As a result, AI-based systems have demonstrated performance comparable to expert radiologists in certain diagnostic tasks. Advanced segmentation and classification techniques play a crucial role in improving detection outcomes. Architectures such as U-Net, DeepLab, and Feature Pyramid Networks are widely used for precise tumor boundary delineation, achieving high segmentation accuracy. EfficientNetB0 has gained attention for its ability to balance performance and computational efficiency through compound scaling, and when combined with FPN, it enhances multi-scale feature learning. Additionally, causal dilated convolutional neural networks improve contextual understanding by capturing long-range spatial dependencies. Hybrid models integrating these techniques have shown superior performance in ovarian cancer detection. However, challenges such as limited annotated datasets, model generalization, and lack of interpretability remain significant barriers, highlighting the need for further research in explainable AI and real-world clinical validation.