Brain tumors arise from abnormal brain cell growth, leading to increased risk of disease and death. MRI scans are crucial for identifying and treating brain cancers by providing images of the brain’s interior. Inaccurate brain tumor segmentation and classification can have fatal consequences. Unfortunately, erroneous brain tumor segmentation and classification results in the loss of many lives from brain tumors. Early detection and accurate diagnosis are paramount. This research introduces a hybrid segmentation technique, Gaussian Hybrid Fuzzy Clustering (GHFC), and an Exponential Cuckoo-based Deep Convolutional Neural Network (Exp. Cuckoo- based DCNN) for brain tumor image classification. The BRATS and SIMBRATS datasets are used to evaluate the proposed method. Our approach surpasses existing state-of-the-art techniques like RBNN, Exp. Cuckoo-based RBNN classifier, and the VGGNet classifier achieving a higher accuracy of 0.98563 and a lower MSE of 0.0472.