基于对比学习的非对称编解码结构的心脏MRI分割研究

doi:10.19678/j.issn.1000-3428.0067980

摘要/Abstract

摘要：

全监督方法在心脏磁共振成像(MRI)分割任务中的成功依赖于大规模标记数据集, 然而由于患者隐私及人工标注困难等问题, 心脏MRI标注数据规模较小, 使全监督方法面临挑战。基于半监督的对比学习方法, 设计双分支编码与单分支解码的心脏MRI分割网络CPCL-Net, 引入图像和像素的联合对比损失, 提升了模型对数据样本的特征表达能力。为了增强CPCL-Net对Hard负样本的分割精度, 设计动态自适应加权模块(DAWM), 利用生成的$ \alpha $和$ \beta $权重因子评估样本级别和像素级别的训练贡献度, 使模型分割精度得到大幅提升。基于自动心脏诊断挑战赛(ACDC)数据集的实验结果表明, 该网络模型仅利用少量标注样本即可获得较高的分割精度, 缓解了心脏MRI高质量标注样本不足导致的分割精度低下问题, 并且在标注样本规模相同的情况下, 对心脏左心室、右心室、心肌等部位的分割精度分别为86.17%、85.52%、84.55%, 优于现有的4种典型半监督分割模型以及经典的对比学习框架Sim-CLR, 有效缓解了全监督分割模型对样本规模的依赖及过拟合问题。

关键词: 对比学习, 动态自适应加权, 医学图像分割, 心脏磁共振成像, 联合损失函数

Abstract:

The success of a fully supervised method in cardiac Magnetic Resonance Imaging (MRI) segmentation tasks relies on large-scale labeled datasets. However, owing to problems associated with patient privacy and difficulty in manual labeling, the small scale of cardiac MRI annotation data makes the fully supervised method challenging. Based on the semisupervised contrastive learning method, a cardiac MRI segmentation network, CPCL-Net, with double-branch coding and single-branch decoding is developed, and the joint contrastive loss of images and pixels is introduced to improve the ability of the model to express the features of the data samples. To enhance the segmentation accuracy of CPCL-Net for Hard negative samples, a Dynamic Adaptive Weighting Module (DAWM) is developed, and the training contribution at the sample and pixel levels is evaluated using the generated and weight factors, which considerably improves the segmentation accuracy of the model. Experimental results based on the Automated Cardiac Diagnosis Challenge (ACDC) dataset indicate that the network model can obtain high segmentation accuracy with only a small number of annotated samples, which alleviates the problem of low segmentation accuracy caused by insufficient high-quality annotated samples of cardiac MRI. The segmentation accuracies for the left ventricle, right ventricle, and myocardium of the heart are 86.17%, 85.52%, and 84.55%, respectively, under the same annotation sample size. The accuracy values are greater than those of the four existing semi-supervised segmentation models and the classical contrastive learning framework, Sim-CLR, which effectively alleviates the dependence and overfitting of the fully supervised model on the sample size.

Key words: contrastive learning, dynamic adaptive weighting, medical image segmentation, cardiac Magnetic Resonance Imaging(MRI), joint loss function

高爽, 史轶伦, 徐巧枝, 于磊. 基于对比学习的非对称编解码结构的心脏MRI分割研究[J]. 计算机工程, 2024, 50(8): 290-300.

Shuang GAO, Yilun SHI, Qiaozhi XU, Lei YU. Research on Cardiac MRI Segmentation Based on Asymmetric Encoding and Decoding Structure of Contrastive Learning[J]. Computer Engineering, 2024, 50(8): 290-300.

https://www.ecice06.com/CN/Y2024/V50/I8/290

图/表 12

图1 对比学习的一般架构

Fig.1 General architecture of contrastive learning

图2 CPCL-Net架构

Fig.2 CPCL-Net architecture

图3 动态自适应加权模型架构

Fig.3 Dynamic adaptive weighting model architecture

图4 T=6时的分割结果

Fig.4 Segmentation results when T=6

图5 对比实验的可视化结果

Fig.5 Visualization results of comparative experiment

图6 动态自适应加权模块有效性验证的可视化结果

Fig.6 Visualization results of dynamic adaptive weighted module effectiveness verification

图7 权重因子

$ \lambda $

分析实验

Fig.7 Analysis experiment of weight factor

$ \lambda $

参考文献 24

1	刘从军, 徐佳陈, 肖志勇, 等. 基于深度学习的心脏核磁共振图像自动分割算法. 计算机工程与科学, 2022, 44(9): 1646- 1654. URL
	LIU C J, XU J C, XIAO Z Y, et al. An automatic cardiac magnetic resonance image segmentation algorithm based on deep learning. Computer Engineering & Science, 2022, 44(9): 1646- 1654. URL
2	HESAMIAN M H, JIA W J, HE X J, et al. Deep learning techniques for medical image segmentation: achievements and challenges. Journal of Digital Imaging, 2019, 32(4): 582- 596. doi: 10.1007/s10278-019-00227-x
3	范馨月, 鲍泓, 潘卫国. 基于类别不平衡数据集的图像实例分割方法. 计算机工程, 2022, 48(12): 224- 231. URL
	FAN X Y, BAO H, PAN W G. Image instance segmentation method based on class-imbalanced dataset. Computer Engineering, 2022, 48(12): 224- 231. URL
4	张重生, 陈杰, 李岐龙, 等. 深度对比学习综述. 自动化学报, 2023, 49(1): 15- 39. URL
	ZHANG C S, CHEN J, LI Q L, et al. Deep contrastive learning: a survey. Acta Automatica Sinica, 2023, 49(1): 15- 39. URL
5	GRILL J B, STRUB F, ALTCHE F, et al. Bootstrap your own latent: a new approach to self-supervised learning[EB/OL]. [2023-06-11]. https://arxiv.org/abs/2006.07733.
6	CHAITANYA K, ERDIL E, KARANI N, et al. Contrastive learning of global and local features for medical image segmentation with limited annotations[EB/OL]. [2023-06-11]. http://arxiv.org/abs/2006.10511.
7	ZENG D W, WU Y W, HU X R, et al. Positional contrastive learning for volumetric medical image segmentation[C]//Proceedings of International Conference on Medical Image Computing and Computer Assisted Intervention. Berlin, Germany: Springer International Publishing, 2021: 221-230.
8	XIANG J P, LI Z W, WANG W J, et al. Self-ensembling contrastive learning for semi-supervised medical image segmentation[EB/OL]. [2023-06-11]. https://arxiv.org/abs/2105.12924.
9	ZHANG S, ZHANG J J, TIAN B, et al. Multi-modal contrastive mutual learning and pseudo-label re-learning for semi-supervised medical image segmentation. Medical Image Analysis, 2023, 83, 102656. doi: 10.1016/j.media.2022.102656
10	SHELHAMER E, LONG J, DARRELL T. Fully convolutional networks for semantic segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4): 640- 651. doi: 10.1109/TPAMI.2016.2572683
11	KHENED M, KOLLERATHU V A, KRISHNAMURTHI G. Fully convolutional multi-scale residual DenseNets for cardiac segmentation and automated cardiac diagnosis using ensemble of classifiers. Medical Image Analysis, 2019, 51, 21- 45. doi: 10.1016/j.media.2018.10.004
12	ZHU H Z, ROHLING R, SALCUDEAN S. Multi-task UNet: jointly boosting saliency prediction and disease classification on chest X-ray images[EB/OL]. [2023-06-11]. http://arxiv.org/abs/2202.07118.
13	WANG W G, ZHOU T F, YU F, et al. Exploring cross-image pixel contrast for semantic segmentation[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2021: 7283-7293.
14	GUTMANN U M, HYVARINEN A. Noise-contrastive estimation: a new estimation principle for unnormalized statistical models[C]//Proceedings of International Conference on Artificial Intelligence and Statistics, Washington D. C., USA: IEEE Press, 2010: 297-304.
15	ZHONG Y Y, YUAN B D, WU H, et al. Pixel contrastive-consistent semi-supervised semantic segmentation[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Washington D. C., USA: IEEE Press, 2021: 7253-7262.
16	付海涛, 刘烁, 冯宇轩, 等. 基于对比学习方法的小样本学习. 吉林大学学报(理学版), 2023, 61(1): 111- 117. URL
	FU H T, LIU S, FENG Y X, et al. Few-shot learning based on contrastive learning method. Journal of Jilin University (Science Edition), 2023, 61(1): 111- 117. URL
17	HE K M, FAN H Q, WU Y X, et al. Momentum contrast for unsupervised visual representation learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2020: 9726-9735.
18	CHANG J H, ZHANG J, XU Y M, et al. Consistency-contrast learning for conceptual coding[C]//Proceedings of the 30th ACM International Conference on Multimedia. New York, USA: ACM Press, 2022: 2681-2690.
19	BERNARD O, LALANDE A, ZOTTI C, et al. Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved?. IEEE Transactions on Medical Imaging, 2018, 37(11): 2514- 2525. doi: 10.1109/TMI.2018.2837502
20	TANABE Y, ISHIDA T, ETO H, et al. Evaluation of the correlation between prostatic displacement and rectal deformation using the Dice similarity coefficient of the rectum[J]. Medical Dosimetry, 2019, 44(4): 39-43.
21	HUTTENLOCHER D P, KLANDERMAN G A, RUCKLIDGE W J. Comparing images using the Hausdorff distance. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1993, 15(9): 850- 863. doi: 10.1109/34.232073
22	PATHAK D, KRÄHENBÜHL P, DONAHUE J, et al. Context encoders: feature learning by inpainting[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Washington D. C., USA: IEEE Press, 2016: 2536-2544.
23	GIDARIS S, SINGH P, KOMODAKIS N. Unsupervised representation learning by predicting image rotations[EB/OL]. [2023-06-11]. http://arxiv.org/abs/1803.07728.
24	ZHANG Y Z, YANG L, CHEN J X, et al. Deep adversarial networks for biomedical image segmentation utilizing unannotated images[M]. Berlin, Germany: Springer International Publishing, 2017.

[1]	郭敏, 张熙涵, 李阳. 融合注意力的教师互一致性半监督医学图像分割[J]. 计算机工程, 2024, 50(9): 313-323.
[2]	钱来, 赵卫伟. 基于对比学习和注意力机制的文本分类方法[J]. 计算机工程, 2024, 50(7): 104-111.
[3]	刘娟, 段友祥, 陆誉翕, 张鲁. 引入知识增强和对比学习的知识图谱补全[J]. 计算机工程, 2024, 50(7): 112-122.
[4]	武星, 殷浩宇, 姚骏峰, 李卫民, 钱权. 面向视频数据的多模态情感分析[J]. 计算机工程, 2024, 50(6): 218-227.
[5]	张宝鑫, 杨丹, 聂铁铮, 寇月. 基于自监督的多视角图协同过滤推荐方法[J]. 计算机工程, 2024, 50(5): 100-110.
[6]	李田芳, 普园媛, 赵征鹏, 徐丹, 钱文华. 基于CLIP和双空间自适应归一化的图像翻译[J]. 计算机工程, 2024, 50(5): 229-240.
[7]	胡帅, 李华玲, 郝德琛. 改进U-Net的多级边缘增强医学图像分割网络[J]. 计算机工程, 2024, 50(4): 286-293.
[8]	李政学, 李枝名, 彭德中, 陈杰. 基于特征对比学习和图卷积的社交网络用户分类[J]. 计算机工程, 2024, 50(4): 258-266.
[9]	张洪程, 李林育, 杨莉, 伞晨峻, 尹春林, 颜冰, 于虹, 张璇. 基于对比学习与语言模型增强嵌入的知识图谱补全[J]. 计算机工程, 2024, 50(4): 168-176.
[10]	吴冠荣, 李元祥, 王艺霖, 陆雨寒, 陈秀华. 基于对比学习的小样本金属表面损伤分类[J]. 计算机工程, 2024, 50(3): 36-43.
[11]	吴永庆, 王钰涵, 朱月. 基于用户多类型反馈行为序列的点击率预估模型[J]. 计算机工程, 2024, 50(10): 405-417.
[12]	曹发鑫, 孙媛媛, 王治政, 潘丁豪, 林鸿飞. 面向借贷案件的相似案例匹配模型[J]. 计算机工程, 2024, 50(1): 306-312.
[13]	顾嘉静, 杨丹, 聂铁铮, 寇月. 基于多视图融合跨层对比学习的推荐算法[J]. 计算机工程, 2024, 50(1): 120-128.
[14]	李晓腾, 张盼盼, 勾智楠, 高凯. 基于多任务学习的多模态命名实体识别方法[J]. 计算机工程, 2023, 49(4): 114-119.
[15]	杨本臣, 贾宇航, 金海波. 融合多分支特征的肝脏和肝脏肿瘤的体积分割[J]. 计算机工程, 2023, 49(10): 194-201.

选择文件类型/文献管理软件名称

选择包含的内容