A novel Grasshopper optimized ResU-Net for Brain Tumor Segmentation

Brain tumors pose significant health risks, making accurate segmentation from MRI images critical for effective diagnosis, treatment planning, and monitoring. Traditional manual segmentation methods are often prone to errors, prompting the development of advanced architectures such as U-Net. However, U-Net faces challenges, including vanishing gradients and a high number of hyperparameters that require careful tuning. To address these limitations, we propose the Grasshopper Optimized ResU-Net model, which incorporates residual blocks within the optimized U-Net framework. This novel architecture enhances brain tumor segmentation by effectively learning complex features from MRI scans while optimizing performance through Grasshopper Optimization. Extensive experimental evaluations demonstrate that the proposed model significantly outperforms existing methods in brain tumor segmentation, highlighting its efficacy in clinical applications.