A Survey of Automatic Brain Tumor Detection and Classification Techniques

Abstract

The timely and accurate detection of brain tumors is a critical challenge in modern healthcare. This survey paper synthesizes recent research on computer-aided automatic brain tumor detection and classification, focusing on methods presented in four contemporary IEEE papers. We analyze the effectiveness of both traditional signal processing (active contour models) and modern deep learning approaches (CNNs like ResNet, EfficientNet, InceptionV3, and VGG-16). The papers are categorized based on their primary methodology, from active contours to deep learning-based detection and classification, with an emphasis on privacy preservation and comprehensive model evaluation. We compare their reported performance metrics, including accuracy, precision, recall, and AUC, to provide a concise overview of the state of the art. The synthesis reveals that deep learning-based approaches, particularly fine-tuned CNN models, consistently achieve high accuracy, while the integration of privacy-preserving techniques is an emerging and vital
research direction.