基于改进SegFormer的胰腺图像分割方法

doi:10.3969/j.issn.1006-2475.2025.06.012

摘要/Abstract

摘要： 摘要：为了解决CT图像中由于胰腺体积小、位置和形状个体差异性较大导致的分割精度不高的问题，本文提出一种基于改进SegFormer模型的胰腺图像分割方法。在模型训练之前，根据胰腺的位置分布来构建候选区域并进行裁剪，从而有效减少背景区域的干扰，降低输入图像分辨率；接着采用SegFormer网络，并引入增大编码分辨率策略，通过调整下采样的比例来增大编码器输出特征图的尺寸，保留更多的细节信息，使模型能更好地应对胰腺的形态变化；然后引入残差极化自注意力模块对编码特征进行通道和空间注意力计算，以突出胰腺区域的关键特征，抑制无关特征的激活，从而提高模型的分割精度。本文方法在NIH数据集上测试的平均DSC为85.5%，参数量和计算量分别为3.91 M和6.89 G FLOPs，表明了该方法在胰腺分割任务上的有效性及其临床应用的潜力。

关键词: 关键词：胰腺分割, SegFormer模型, 极化自注意力, 轻量级, CT

Abstract: Abstract: To address the issue of low segmentation accuracy in CT images due to the small volume of the pancreas and significant individual differences in its position and shape， we propose an improved SegFormer-based method for pancreatic image segmentation. Prior to model training， we construct candidate regions based on the distribution of the pancreas and perform cropping to effectively reduce background interference and lower the input image resolution. Next， we employ the SegFormer network and introduce an encoding resolution enhancement strategy by adjusting the downsampling ratio to increase the size of the encoder's output feature maps， which retains more detail information to better handle morphological variations of the pancreas. We then incorporate residual polarized self-attention modules to compute channel and spatial attention on the encoded features， highlighting key characteristics of the pancreatic region while suppressing the activation of irrelevant features， thus improving the model's segmentation accuracy. The proposed method achieved an average DSC of 85.5% on the NIH dataset， with a parameter count of 3.91 M and a computational load of 6.89 G FLOPs， indicating its effectiveness in the pancreatic segmentation task and its potential for clinical applications.

Key words: Key words: pancreas segmentation, SegFormer, polarized self-attention, lightweight, CT

中图分类号:

TP391

梁攀如, 辛国江, 丁长松. 基于改进SegFormer的胰腺图像分割方法[J]. 计算机与现代化, 2025, 0(06): 71-78.

LIANG Panru, XIN Guojiang, DING Changsong. Pancreatic Image Segmentation Approach Based on Improved SegFormer[J]. Computer and Modernization, 2025, 0(06): 71-78.

参考文献

［1］ SIEGEL R L， MILLER K D， JEMAL A. Cancer statistics， 2019［J］. CA: A Cancer Journal for Clinicians， 2019，69（1）:7-34.
［2］ SUNG H， FERLAY J， SIEGEL R L， et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries［J］. CA: A Cancer Journal for Clinicians， 2021，71（3）:209-249.
［3］ LIU S Q， LIANG S J， HUANG X， et al. Graph-enhanced U-Net for semi-supervised segmentation of pancreas from abdomen CT scan［J］. Physics in Medicine & Biology， 2022，67（15）. DOI: 10.1088/1361-6560/ac80e4.
［4］ CHU L C， GOGGINS M G， FISHMAN E K. Diagnosis and detection of pancreatic cancer［J］. The Cancer Journal， 2017，23（6）:333-342.
［5］ KARASAWA K， ODA M， KITASAKA T， et al. Multi-atlas pancreas segmentation: Atlas selection based on vessel structure［J］. Medical Image Analysis， 2017，39:18-28.
［6］ TAM T D， BINH N T. Efficient pancreas segmentation in computed tomography based on region-growing［C］// Nature of Computation and Communication: International Conference. Springer， 2015:332-340.
［7］ FARAG A， LU L， TURKBEY B， et al. A bottom-up approach for automatic pancreas segmentation in abdominal CT scans［C］// Abdominal Imaging. Computational and Clinical Applications: 6th International Workshop. ABDI 2014， Held in Conjunction with MICCAI 2014， Springer， 2014: 103-113.
［8］曹路洋，李建微. 面向深度学习的胰腺医学图像分割方法研究进展［J］.小型微型计算机系统， 2022，43（12）:2591-2604.
［9］ ZHOU Y Y， XIE L X， SHEN W， et al. A fixed-point model for pancreas segmentation in abdominal CT scans［C］// International Conference on Medical Image Computing and Computer Assisted Intervention. Springer， 2017:693-701.
［10］ XIE L X， YU Q H， ZHOU Y Y， et al. Recurrent saliency transformation network for tiny target segmentation in abdominal CT scans［J］.IEEE Transactions on Medical Imaging， 2019，39（2）:514-525
［11］ OKTAY O， SCHLEMPER J， FOLGOC L L， et al. Attention u-net: Learning where to look for the pancreas［J］. arXiv preprint arXiv:1804.03999， 2018.
［12］ RONNEBERGER O， FISCHER P， BROX T. U-Net: Convolutional networks for biomedical image segmentation［C］//Medical Image Computing and Computer-Assisted Intervention 18th International Conference. Springer， 2015:234-241.
［13］ LI M Y， LIAN F H， GUO S X. Pancreas segmentation based on an adversarial model under two-tier constraints［J］.Physics in Medicine & Biology，2020，65（22）. DOI: 10.1088/1361-6560/abb6bf.
［14］ LI M， LIAN F， WANG C， et al. Accurate pancreas segmentation using multi-level pyramidal pooling residual U-Net with adversarial mechanism［J］.BMC Medical Imaging，2021，21（1）:1-8.
［15］毕秀丽，陆猛，肖斌，等. 基于双解码U型卷积神经网络的胰腺分割［J］. 软件学报， 2022，33（5）:1947-1958.
［16］张国栋，唐晓艺，鞠蓉晖，等.融合双注意力机制的多尺度胰腺分割方法［J］. 计算机工程与设计， 2024，45（4）:1189-1194.
［17］ DOSOVITSKIY A， BEYER L， KOLESNIKOV A， et al. An image is worth 16x16 words: Transformers for image recognition at scale［J］.arXiv preprint arXiv:2010.11929，2020.
［18］ VASWANI A， SHAZEER N， PARMAR N， et al. Attention is all you need［J］. avXiv preprint avXiv: 1706.03762， 42017.
［19］ CHEN J， LU Y Y， YU Q H， et al. Transunet: Transformers make strong encoders for medical image segmentation［J］.arXiv preprint arXiv:2102.04306， 2021.
［20］ XIE E Z，WANG W H，YU Z D，et al. SegFormer: Simple and efficient design for semantic segmentation with transformers［J］. Advances in Neural Information Processing Systems， 2021，34:12077-12090.
［21］ LIU H J， LIU F Y， FAN X Q， et al. Polarized self-attention: Towards high-quality pixel-wise regression［J］. arxiv preprint arxiv:2107.00782， 2021.
［22］ CHEN H P， LIU Y J， SHI Z A， et al. Pancreas segmentation by two-view feature learning and multi-scale supervision［J］.Biomedical Signal Processing and Control，2022， 74. DOI: 10.1016/j.bspc.2022.103519.
［23］朱纷，何立风，孙爽，等.基于形变残差和级联编码的胰腺分割模型［J］. 计算机与现代化， 2024，（6）:83-88.
［24］ LIU Z， SU J， WANG R H， et al. Pancreas co-segmentation based on dynamic ROI extraction and VGGU-Net［J］. Expert Systems with Applications， 2022，192. DOI: 10.1016/j.eswa.2021.116444.
［25］ ROTH H R， LU L， FARAG A， et al. Deeporgan: Multi-level deep convolutional networks for automated pancreas segmentation［C］// Medical Image Computing and Computer-Assisted Intervention 18th International Conference， Springer， 2015:
556-564.
［26］ LIN T Y， GOYAL P， GIRSHICK R， et al. Focal loss for dense object detection［C］// Proceedings of the 2017 IEEE International Conference on Computer Vision，2017:2980-2988.
［27］ ZHAO H S， SHI J P， QI X J， et al. Pyramid scene parsing network［C］// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2017: 2881-2890.
［28］ CHEN L C， ZHU Y K， PAPANDREOU G， et al. Encoder-decoder with atrous separable convolution for semantic image segmentation［C］// Proceedings of the European Conference on Computer Vision （ECCV）. 2018:801-818.
［29］ SUN K， XIAO B， LIU D， et al. Deep high-resolution representation learning for human pose estimation［C］// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019:5693-5703.

[1]	时现伟1, 范鑫2. 基于轻量化的视频帧场景语义分割方法[J]. 计算机与现代化, 2024, 0(08): 49-53.
[2]	刘静乐, 罗翔, 宫成荣, 张国鹏. 基于RF-RFECV和LightGBM算法的糖尿病预测[J]. 计算机与现代化, 2023, 0(11): 36-43.
[3]	刘付谦, 秦华妮, 赖惠慧. 基于SMOTE和贝叶斯优化的Adj-LightGBM人岗匹配算法[J]. 计算机与现代化, 2023, 0(03): 90-95.
[4]	吴丽君, 蔡周威, 陈志聪. 基于改进残差特征蒸馏的轻量级超分辨率网络[J]. 计算机与现代化, 2022, 0(11): 89-94.
[5]	李燕, 卢峥松, 李青云, 杨世海, 张小龙. 基于轻量级结构重参数化网络的口罩检测算法[J]. 计算机与现代化, 2022, 0(07): 40-46.
[6]	梁超, 黄洪全. 基于卷积神经网络的轻量级图像超分辨率[J]. 计算机与现代化, 2020, 0(11): 23-27.
[7]	简碧园1, 汪海涛2, 刘道微1. 一种RFID双向认证协议[J]. 计算机与现代化, 2017, 0(7): 107-110.
[8]	郭峰,曾培峰. 基于分布式服务器的视频监视系统[J]. 计算机与现代化, 2015, 0(6): 74-77+100.
[9]	张连福. 基于J2EE轻量级框架的人力资源管理系统的研究与实现[J]. 计算机与现代化, 2012, 1(10): 148-150，.
[10]	孙立;李昕. 基于SSi2框架处理LOB数据的方法研究[J]. 计算机与现代化, 2010, 1(6): 183-0186.
[11]	史笛扬;贾卓生. J2EE轻量级框架在高校课程管理系统中的应用[J]. 计算机与现代化, 2009, 8(8): 152-156.