Augmenting U-Net with Attention for Smarter Tumor Detection in MRI Scans

In neuro-oncology, precise and effective brain tumor detection using magnetic resonance imaging (MRI) is essential for finding therapy planning, and follow-up. To increase the accuracy and dependability of brain tumor detection, this paper presents an augmented U-Net architecture that has been improved with attention mechanisms. Utilizing attention gates and SE blocks as enhancements to the standard U-Net design, the model suppresses extraneous structures and background noise while selectively focusing on pertinent tumor locations. When tested on benchmark datasets like BraTS, the suggested model outperforms baseline U-Net and other cutting-edge models in provisos of detection accurateness, sensitivity, and Dice Similarity Coefficient (DSC). In addition to improving spatial feature learning, the attention-augmented U-Net guarantees resilience across various tumor types and MRI modalities. The implementation of intelligent, automated tools in clinical processes for brain tumor analysis has advanced significantly as a result of this work.