U-Net-based approaches for brain tumor segmentation

Brain tumor segmentation is crucial for diagnosis and treatment planning. Convolutional neural networks (CNNs) have shown potential in medical image analysis, with U-Net architecture emerging as a leading technique for segmentation. The chapter comprehensively explores the implementation and evaluation of CNNs in magnetic resonance imaging (MRI) images, including an extensive literature review covering the background of brain tumors, MRI, and radiologists' role in image analysis. Results showed that high-quality segmentation can be achieved with simpler models and fewer computational resources, making it possible for researchers and practitioners with limited resources to develop effective models without resource availability constraints. This chapter provides valuable insights and guidance for future research in this area. The Residual U-Network architecture emerged as the best-performing model, achieving promising results despite limited resources and data during training. Future work should refine and explore additional modifications to improve segmentation performance.