MA2-Net: Multi-Scale Adaptive Attention-Enhanced Network for Brain Tumor Detection

Medical image segmentation is crucial for the diagnosis and treatment of various diseases, particularly in brain tumor detection. In this paper, we present MA2-Net, a novel deep learning architecture specifically designed for medical image segmentation tasks. MA2-Net incorporates three key modules: the Channel-Prioritized Multi-Scale Attention Downsampling Module, the Multi-Scale Adaptive Spatial Attention Gate, and the Efficient Up-Convolution Block. These modules work collaboratively to improve feature extraction, enhance computational efficiency, and increase segmentation accuracy, effectively addressing the limitations of traditional U-Net models. Experimental results demonstrate that MA2-Net significantly outperforms existing methods, achieving higher accuracy with minimal computational cost, making it an ideal solution for resource-constrained environments.