📄 中文摘要
准确的息肉分割在结肠镜检查中对癌症预防至关重要，但由于形态变异性大、与正常结构（如褶皱和血管）视觉相似度高以及需要稳健的多尺度检测等原因，仍然面临挑战。现有的深度学习方法存在单向处理、弱多尺度融合和缺乏解剖约束的问题，常导致假阳性（正常结构的过度分割）和假阴性（未能检测到微妙的平坦病变）。GRAFNet是一种生物启发的架构，模拟人类视觉系统的层次组织，集成了三个关键模块，以提升息肉分割的准确性和鲁棒性。
GRAFNet：通过引导皮层注意反馈实现多尺度视网膜处理以增强医学图像息肉分割
📄 English Summary
GRAFNet: Multiscale Retinal Processing via Guided Cortical Attention Feedback for Enhancing Medical Image Polyp Segmentation
Accurate polyp segmentation in colonoscopy is crucial for cancer prevention but remains challenging due to high morphological variability, strong visual similarity to normal structures such as folds and vessels, and the necessity for robust multi-scale detection. Existing deep learning approaches suffer from unidirectional processing, weak multi-scale fusion, and the lack of anatomical constraints, often resulting in false positives (over-segmentation of normal structures) and false negatives (missed subtle flat lesions). GRAFNet is proposed as a biologically inspired architecture that emulates the hierarchical organization of the human visual system. It integrates three key modules to enhance the accuracy and robustness of polyp segmentation.