Abstract
ABSTRACT
Uterine cancer is one of the most common malignancies in women, and early detection is vital for improving clinical outcomes and enhancing women's well‐being. However, subtle tissue variations make manual diagnosis subjective and time‐consuming. This study presents a transformer‐enhanced deep convolutional neural network (TE‐DCNN) for automated classification and segmentation of uterine histopathological images. The architecture integrates convolutional layers for local feature extraction with transformer modules for global contextual learning, supported by dual channel–spatial attention (SA) to highlight diagnostically critical regions. Evaluated on four classes—Normal Endometrium, Endometrial Hyperplasia, Endometrial Polyp, and Endometrial Adenocarcinoma, the model achieved 99% accuracy, 0.97 sensitivity, and 0.98 specificity, outperforming CNN and vision transformer (ViT) baselines. Visual segmentation confirmed precise localization of malignant areas, enhancing interpretability. The proposed TE‐DCNN offers a fast, explainable, and reliable tool for early uterine cancer diagnosis and can be extended to other gynecological imaging tasks.