Deep Learning: Techniques for Ovarian Cancer Subtype Classification and Outlier Detection (Deep Learning: Técnicas para Classificação de Subtipos de Câncer Ovariano e Detecção de Outliers)

This study investigates the application of deep neural networks for ovarian cancer subtype classification and outlier detection, aiming to improve diagnostic accuracy and address interobserver variability. Using the OCEAN Dataset and advanced deep learning techniques, EfficientNetV2 and ResNet152 architectures were compared. The findings demonstrate the superior performance of ResNet152, especially in rare subtypes, underscoring its potential for precise and scalable clinical diagnostics. This work enhances clinical practice by offering a scalable and accurate diagnostic support model.