Advancing Brain Tumor Management Using Generative AI: A Comprehensive Review of MRI-Based Applications

Abstract

Purpose: To systematically review the integration of generative artificial intelligence (GenAI) in MRI-based brain tumor management and evaluate its impact on detection, segmentation, diagnosis, treatment planning, and outcome prediction. The review addresses the research question like How effectively can GenAI enhance MRI-based brain tumor management, and what challenges limit its clinical translation. Methods: A systematic literature review was conducted following PRISMA guidelines across PubMed, IEEE Xplore, ScienceDirect, SpringerLink, and Web of Science. Studies published between 2015 and 2025 that applied generative models—including Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), diffusion models, and transformer-based methods to brain tumor MRI analysis were included. Data were extracted from 55 peer-reviewed studies for qualitative synthesis. Results: Hybrid architectures combining GANs with CNNs or transformers achieved classification accuracies up to 99.53%. Attention-based segmentation models on multi-modal MRI data attained Dice scores exceeding 0.89. GenAI demonstrated significant contributions to tumor detection, cross-modal synthesis, data augmentation, treatment personalization, and survival prediction. Key challenges included data scarcity, model explainability, algorithmic bias, and the need for prospective clinical validation. Conclusions: GenAI offers transformative potential for neuro-oncological care by addressing data limitations and enabling personalized treatment strategies. However, resolving ethical concerns, ensuring regulatory compliance, and advancing multi-modal integration remain essential for successful clinical translation