Novel methods and materials are used in healthcare applications for finding cancer in various parts of the human system. To select the most suitable therapy plan for individuals with domestically progressed cervical cancer, robustness metrics are required to estimate their early phase. The goal of the research is to increase the effectiveness of cervical cancer patients' detection by using deep learning-based radiomics assessment of magnetic resonance imaging (MRI). From March 2016 to November 2019, 125 patients with early-stage cervical cancer provided 980 dynamic X1 contrast-enhanced (X1DCE) and 850 X2 weighted imaging (X2WI) MRI images for training and testing. A convolutional neural network model was used to estimate cervical cancer state based on the specified characteristics. The X1DCE exhibited high discriminative outcomes than X2WI MRI in terms of prediction ability, as calculated by the confusion matrix assessment and receiver operating characteristic (ROC) curve approach. The mean maximum region under the curve of 0.95 was found using an attentive ensemble learning method that included both MRI sequencing (Sensitivity = 0.94, Specificity = 0.94, and accuracy = 0.96). Whenever compared with conventional radiomic approaches, the results show that a variety of radiomics based on deep learning might be created to help radiologists anticipate vascular invasion in patients with cervical cancer before surgery. Based on radiomics technique, it has proven to be an effective tool for estimating cervical cancer in its early stages. It would help people choose the best therapy method for them and make medical judgments. © 2024 Elsevier B.V., All rights reserved.