Classifying brain tumors is a critical aspect of medical image analysis, significantly enhancing the precision of diagnosis and therapy planning. In recent years, deep learning methodologies have exhibited considerable success in several medical image analysis applications, including brain tumor classification. This research presents a deep learning-based methodology for the categorization of multi-class brain tumors using MRI data. To standardize the classification of different tumor types, this method use convolutional neural networks (CNNs) to autonomously extract discriminative characteristics from MRI data. The methodology employs MRI scans of patients with pituitary tumors, meningiomas, and gliomas sourced from the FigShare dataset. The thorough testing and assessment confirm the efficacy of our approach to accurate brain tumor classification across multiple categories. The accuracy, sensitivity, and specificity of the results produced by the proposed model are remarkable for different tumor types, demonstrating the potential of the CNN (ResNet50) architecture to enhance diagnostic processes in neuroimaging. Furthermore, it provides critical insights into the characteristics of the acquired model and the therapeutic relevance of the results. This research enhances the knowledge concerning the utilization of advanced neural network methodologies to improve brain tumor diagnosis and emphasizes the significance of automated classification systems in clinical practice.