Publication

Abstract : Endometrial cancer [17], one of the most common gynecologic malignancies, requires accurate classification for proper treatment planning. The conventional diagnostic methods, such as endometrial biopsy by dilation and curettage or hysteroscopy, are invasive and may not be very accurate. Breakthroughs in AI and DL [16] bring about revolutionary solutions to improve the classification of endometrial cancer. This paper reviews current DL models such as SWIN, ResNet, and DenseNet for hysteroscopic and MR imaging, achieving high accuracy and computational efficiency for clinical applications. Ensemble methods, and automated MR segmentation can improve classification and tumor analysis. Emerging explainable AI techniques enhance transparency, which integrates with hysteroscopy facilitating real-time classification and reducing the invasive procedures. Future research will focus on compact systems, multicenter validation, and richer datasets to ensure strong performance across demographics and disease stages.