基于注意力门控的 U-Net 模型用于脑肿瘤的语义分割及生存预后特征提取

Gliomas, among the most common primary brain tumors, exhibit significant variability in aggressiveness, prognosis, and histology, posing challenges for treatment due to complex and time-consuming surgical interventions. A novel Attention-Gated Recurrent Residual U-Net (R2U-Net) based Triplanar (2.5D) model is proposed to enhance brain tumor segmentation. The model integrates residual, recurrent, and triplanar architectures to improve feature representation and segmentation accuracy while maintaining computational efficiency, potentially aiding in better treatment planning. The proposed method achieves a Dice Similarity Score (DSC) of 0.900 for Whole Tumor (WT) segmentation on the BraTS2021 validation set, demonstrating performance comparable to leading approaches.