Brain tumors are among the most critical neurological disorders, significantly affecting patients’ health and prognosis. Early detection and accurate classification play a vital role in the treatment process. Although Magnetic Resonance Imaging (MRI) is a widely used noninvasive imaging modality, classification accuracy remains limited due to the morphological similarities among tumor types and inconsistencies in image quality. Deep learning models have shown remarkable success in medical image analysis; however, individual models often lack robustness and may suffer performance degradation when dealing with complex or atypical data. This paper proposes Fuzzy-EDTrans, an ensemble deep learning model that integrates three advanced architectures: EfficientNet, Vision Transformer, and DenseNet, using the Fuzzy Sugeno Integral to optimize classification accuracy for brain tumors. This approach not only leverages the individual strengths of each architecture but also enhances the aggregation of information and improves decisionmaking precision. Moreover, it increases the model’s stability when processing complex medical images. Experiments conducted on the public BrTC2020 dataset demonstrate that the proposed model achieves a classification accuracy of 97.91%, outperforming standalone models. These results confirm the potential of the proposed approach in supporting AI-based medi-cal image diagnosis, especially in the development of intelligent healthcare systems.