Brain tumor segmentation using a combination of graph cut and optimized U-Net with genetic algorithm: An efficient hybrid framework for enhanced medical image analysis

Accurate segmentation of low-grade gliomas (LGG) in FLAIR MRI remains challenging due to low-contrast tumor boundaries, irregular morphology, and extreme class imbalance. This work presents a lightweight hybrid segmentation framework that integrates three complementary components: (1) a fast graph-based region refinement step (using GrabCut as an efficient approximation of graph-cut segmentation) applied during training only to enhance tumor-background contrast, (2) a compact U-Net architecture whose filter configuration is automatically selected using a small-scale genetic algorithm (GA), and (3) an analytical evaluation of lightweight post-processing strategies, studied separately from the final GA-optimized model. The method was evaluated on the LGG-MRI-Segmentation dataset using strict patient-level splits (79/19/12). The final model achieved strong slice-wise performance with a mean Dice of 0.9132 and an exceptionally high median Dice of 0.9855, demonstrating both robustness to challenging cases and near-perfect agreement on the majority of slices. Additional metrics include IoU = 0.8851, sensitivity = 0.8954, precision = 0.9523, specificity = 0.9997, and HD95 = 1.00 mm, indicating highly accurate boundary delineation. For a fair benchmark, a 2D nnU-Net was trained on the same dataset and patient-level splits. nnU-Net achieved a mean validation Dice of 0.8450 and a best EMA Dice of 0.9134. The proposed model therefore matches nnU-Net in mean Dice while substantially outperforming it in median Dice and achieving a markedly lower HD95, all with 90% less training time and significantly reduced computational cost. Taken together, these findings suggest that a well-coordinated combination of preprocessing, lightweight architectural tuning, and efficient post-processing can yield performance competitive with large automated frameworks while remaining computationally inexpensive. This makes the proposed system well-suited for real-time clinical deployment and resource-constrained environments.