Brain tumor diagnosis from magnetic resonance imaging (MRI) remains a challenging task due to the presence of noise, complex tumor structures, and high-dimensional image features. Accurate detection and classification are essential for effective clinical decision-making and treatment planning. To address these challenges, this study proposes a novel RGCS-NCGNN-AAM framework for automated brain tumor analysis. The proposed approach integrates advanced preprocessing, dimensionality reduction, segmentation, classification, and optimization techniques into a unified pipeline. Initially, MRI images are enhanced using a LABF-based preprocessing method to suppress noise while preserving important structural information. Subsequently, Spatially Driven Principal Component Analysis (SD-PCA) is employed to reduce feature dimensionality and retain discriminative tumor-related characteristics. For precise tumor localization, a LA-ST segmentation technique is introduced to accurately delineate tumor boundaries and improve region extraction. The segmented features are then processed through a Robust Graph Convolution Structure-based Neural Context Graph Neural Network (RGCS-NCGNN), which effectively captures both topological and semantic relationships among data samples. Furthermore, an Adaptive Altruistic Mutualism (AAM) optimization strategy is incorporated to dynamically adjust model parameters and enhance learning performance. The proposed framework was evaluated using benchmark brain tumor MRI datasets and demonstrated superior performance in terms of classification accuracy, segmentation quality, and model robustness. Experimental results indicate that the integration of graph-based learning and adaptive optimization significantly improves diagnostic reliability compared with conventional machine learning and deep learning approaches. In addition, the framework preserves critical spatial information, enabling more accurate tumor characterization and boundary identification. The proposed method also exhibits strong scalability and generalization capabilities across diverse MRI samples. These findings suggest that the RGCS-NCGNN-AAM framework can serve as an effective computer-aided diagnostic tool for supporting radiologists and healthcare professionals in brain tumor assessment. Future research will focus on incorporating quantum graph neural networks and self-supervised learning techniques to further enhance computational efficiency and diagnostic performance in real-world clinical environments.