<p>Endometrial cancer (EC) ranks as the sixth most frequently observed cancer among women worldwide. The proposed study introduces an efficient computer aided diagnosis (CAD) system that can assist physicians or healthcare providers to investigate, estimate, and classify endometrial cancer using histopathological images. The methodology involves extraction of handcrafted and deep learning features, noise removal using haze noise removal technique, Vahadane stain normalization, segmentation using K means clustering, and classification. Handcrafted features relying on colour layout, texture, edges, and HSV are extracted post-enhancement and noise removal, whereas deep learning features are extracted using InceptionV3 network from the segmented, normalized images. Heterogeneous feature vector dataset generated using middle-level fusion of handcrafted and deep features is further processed using MLP classifier. The proposed model outperforms various state-of-the-art methods achieving an accuracy, precision, and F1 score of approximately 97.40%. The model’s ability to offer enhanced classification of endometrial cancer is further confirmed by AUC-ROC curves, minimal false alarms (0.90%), and 4% miss rate.</p>