MRI diagnosis and classification of brain tumors are crucial for early diagnosis and effective treatment. This paper presents a deep and machine learning-based approach to improve the accuracy and reliability of detecting brain tumors. A dataset consisting of 255 MRI images was utilized, and the advanced preprocessing techniques involved adaptive contrast enhancement, noise reduction, and segmentation using fuzzy logic to prepare data. The feature extracted methods involved the enrichment of Gray Level Co-occurrence Matrix in a dataset. The proposed EDN-SVM detection had an accuracy of 97.79% for EDN-SVM and outperformed previous techniques. Such a robust nature of this model was again validated on various metrics with regard to sensitivity, specificity, Jaccard Coefficient, and demonstrated great clinical application possibilities in supporting the radiologist to spot correct and time-efficient diagnosis.