Image segmentation in medical imaging is difficult. The axil, corneal, and sagittal views of MRI may assist clinicians diagnose human brain tumors. Rain tumors are more invasive, making MR imaging difficult. By binarizing and deleting unneeded elements, level set techniques and human participation generated a discontinuous contour that highlighted the sick human brain tumor component. This research presents a new DWT-PCA architecture for MRI image classification. Our objective is to categorize MRI images into benign and malignant categories to improve tumor segmentation and detection. For example, DWT is performed on the tumor to discover key features. We use multi-level DWT to collect image characteristics. PCA reduces feature space while retaining the most significant DWT coefficient discriminating data. An SVM classifier receives normalized contrast, correlation, energy, homogeneity, mean, standard deviation, entropy, RMS, variance, smoothness, kurtosis, skewness, and inverse difference moment. SVM classifiers are trained on benign and malignant brain MRI images. Our findings show that the recommended technique successfully classifies MRI images. The framework has high classification accuracy and tumor detection and segmentation capability. Cross-validation and hold-out approaches verify classification accuracy, proving our system’s resilience.