Brain tumor segmentation using deep learning: A Review
Year: 2025
Type: review
Abstract: Brain tumor segmentation is a crucial task in medical image analysis, as accurate delineation of tumor regions is vital for clinical diagnosis, treatment planning, and prognosis assessment. Traditional Convolutional Neural Network (CNN)-based models have demonstrated significant success in capturing local features, but they face challenges in modeling global context, which is essential for complex segmentation tasks. This review examines recent advancements in brain tumor segmentation, with a focus on CNNs, Transformers, Mamba, and Graph Neural Networks (GNNs), as well as their hybrid models. This review critically evaluates the strengths and limitations of each approach with respect to architecture, segmentation accuracy, and real-world applicability. Additionally, it addresses key challenges such as computational complexity and data scarcity, and proposes future research directions to enhance the practical use of these methods in clinical settings.
DOI: 10.54097/31ag9n29
Authors: Beibei Hou, Saizong Guan
Language: English
Subfield: Neurology
Field: Neuroscience
Domain: Life Sciences
Open Access status: hybrid
OpenAlex ID: W4407973826