Retrospectively collect preoperative transvaginal B-mode ultrasound (BMUS), color Doppler flow imaging (CDFI) and three-dimensional ultrasound (3D-US) images and clinical data in patients with non-endometrial cancer diseases and endometrial cancer confirmed by pathology. They were grouped as training set(Tongji Hospital of Tongji Medical College, Huazhong University of Science and Technology) and external validation set(Women's Hospital, School of Medicine, Zhejiang University) . Radiomics features were extracted from corresponding transvaginal ultrasound images. Then, the minimum redundancy maximum relevance (mRMR) algorithm and the least absolute shrinkage and selection operator (LASSO) regression were used to select the non- malignant or malignant status-related features and cervical stromal invasion (CSI) status or non-CSI status features and construct the transvaginal ultrasound radiomics score (Rad-score). Multivariate logistic regression was performed using the three radiomics score together with clinical data, and subsequently develop a nomogram to diagnosis endometrial cancer and CSI respectively. The performance of the nomogram was assessed by discrimination, calibration, and clinical usefulness in the training and external validation set.