[论文解读] Adaptive Context Selection for Polyp Segmentation

Ruifei Zhang, Guanbin Li+4 more|arXiv (Cornell University)|Jan 12, 2023

Radiomics and Machine Learning in Medical Imaging被引用 12

一句话总结

本文提出 ACSNet——一个带有局部上下文注意力（LCA）、全局上下文模块（GCM）和自适应选择模块（ASM）的编码器–解码器框架，能够自适应融合局部与全局上下文用于息肉分割，在 EndoScene 和 Kvasir-SEG 数据集上实现了最先进的结果。

ABSTRACT

Accurate polyp segmentation is of great significance for the diagnosis and treatment of colorectal cancer. However, it has always been very challenging due to the diverse shape and size of polyp. In recent years, state-of-the-art methods have achieved significant breakthroughs in this task with the help of deep convolutional neural networks. However, few algorithms explicitly consider the impact of the size and shape of the polyp and the complex spatial context on the segmentation performance, which results in the algorithms still being powerless for complex samples. In fact, segmentation of polyps of different sizes relies on different local and global contextual information for regional contrast reasoning. To tackle these issues, we propose an adaptive context selection based encoder-decoder framework which is composed of Local Context Attention (LCA) module, Global Context Module (GCM) and Adaptive Selection Module (ASM). Specifically, LCA modules deliver local context features from encoder layers to decoder layers, enhancing the attention to the hard region which is determined by the prediction map of previous layer. GCM aims to further explore the global context features and send to the decoder layers. ASM is used for adaptive selection and aggregation of context features through channel-wise attention. Our proposed approach is evaluated on the EndoScene and Kvasir-SEG Datasets, and shows outstanding performance compared with other state-of-the-art methods. The code is available at https://github.com/ReaFly/ACSNet.

研究动机与目标

• 通过解决息肉大小和形状多样性来推动准确的息肉分割。
• 开发一个编码器–解码器框架，选择性地利用局部与全局上下文。
• 引入模块（LCA、GCM、ASM）以基于息肉大小自适应融合上下文特征。
• 在公开数据集（EndoScene、Kvasir-SEG）上评估并展示相对于现有方法的性能提升。

提出的方法

• 具有五块 ResNet34 编码器/解码器的增强 UNet-like 编码器–解码器。
• Local Context Attention（LCA）模块取代跳跃连接，通过前一预测图引导的局部区域的硬区域上下文，丰富解码器特征。
• Global Context Module（GCM）通过池化分支和非局部操作捕获多尺度全局上下文，并将上下文传递给所有 ASM 块。
• Adaptive Selection Module（ASM）使用 squeez–and–excitation 风格的通道注意力，将局部与全局上下文与先前解码器特征进行融合。
• 在解码器的多个阶段对预测图进行监督学习；损失由二元交叉熵和 Dice 损失组合而成。

实验结果

研究问题

• RQ1 自适应选择局部与全局上下文是否能在不同大小和形状的息肉分割中实现改进？
• RQ2 LCA、GCM 和 ASM 分别以及共同对分割精度有何贡献？
• RQ3 ACSNet 是否在标准基准测试上相较于现有架构实现一致的改进？

主要发现

• ACSNet 在 EndoScene 的 Dice 得分为 86.59%，为所列方法中最佳，相比第二名提升 3.66%。
• 在 EndoScene，ACSNet 还达到 Rec 87.96，Spec 99.16，Prec 90.99，IoUp 79.73，IoUb 96.86，mIoU 88.29，Acc 97.11。
• 在 Kvasir-SEG，ACSNet 达到 Dice 91.30 和 mIoU 90.28，优于 UNet、UNet++、SFANet 和 SegNet 变体；Baseline+LCAs+GCM 及其他消融也显示 Progressive gains。
• 消融研究表明，逐步添加 LCAs、GCM 和 ASM，Dice 依次提升 0.79%、1.28% 和 1.02% 相对于基线。