【摘要】 在医学图像处理中，息肉分割面临息肉形态变异、边界模糊等情况.因此，本文提出通道增强注意力模块(CEAM)与多尺度特征融合模块(MFFM).CEAM强化编码层与解码层特征交互，由学习注意力权重增强特征图相关通道，提升细粒度特征表征能力；MFFM融合不同扩张率分支特征，同步捕获息肉图像细节与全局信息，适配不同尺寸息肉特征处理.本文在4个公开息肉数据集进行大量实验，结果验证所提出方法的分割效果更佳，性能优异.更多还原

【Abstract】 This paper proposes a channel enhancement attention module(CEAM).It enhances the feature interactions between the encoding and decoding layers and boosts the representation capacity of fine-grained features by learning attention weights to enhance relevant channels in the feature maps.Furthermore, this paper introduces a multi-scale feature fusion module(MFFM),which fuses feature information from branches with different dilation rates to capture both the detailed and global information of polyp images, enabling the model to handle polyp features of various sizes.Finally, the proposed method is extensively evaluated on four publicly available polyp datasets.Experimental results demonstrate that the proposed method achieves superior segmentation performance with excellent performance.更多还原