基于双尺度纠缠互补学习的半监督医学图像分割算法

doi:10.13229/j.cnki.jdxbgxb.20260106

Abstract

Abstract:

Semi-supervised learning serves as a core technique to address the scarcity of labeled data in medical image segmentation. However， existing methods struggle to achieve satisfactory segmentation results for complex medical images with high morphological heterogeneity and blurred boundaries. This paper proposes a semi-supervised segmentation algorithm based on Dual-Scale Entangled Complementary Learning. A Dual-Scale Entangled Complementary Decoder is constructed to establish collaborative complementary relationships between dual-scale features and realize complementary modeling of structural information. Specifically， the Cross-level Fusion Encoder Module strengthens the interaction between shallow detailed features and deep semantic features， while the Multi-Dimensional Feature Optimization Module improves the reconstruction capability of spatial structures and boundary details during decoding. Supervised loss together with perturbation and scale consistency losses are adopted to implement semi-supervised training. Experimental results demonstrate that the proposed algorithm achieves competitive segmentation performance on datasets of colon polyps， skin lesions and brain tumors， and maintains favorable stability in segmentation accuracy， boundary smoothness and cross-test-set generalization ability.

Key words: medical image segmentation, semi-supervised learning, entangled complementary decoder, cross-level fusion encoder, multi-dimensional feature optimization

CLC Number:

TP391

Xue WANG,Yong-fa ZHU,Hong-yu LI,Wei-peng ZHOU,Tai-hui LIU. Semi-supervised medical image segmentation algorithm based on dual-scale entangled complementary learning[J].Journal of Jilin University(Engineering and Technology Edition), 2026, 56(10): 2695-2705.

Figures/Tables 12

Fig.1

Fig. 2

Fig.3

Table 1

Comparison results on CVC-ClinicDB and Kvasir"

数据集	方法	评价指标
数据集	方法	mDice	mIoU	$F β ω$	$S α$	MAE
CVC-ClinicDB	URPC^［25］	0.781	0.692	0.777	0.823	0.026
	SSNet^［26］	0.809	0.712	0.785	0.858	0.023
	MCNet^［27］	0.796	0.714	0.780	0.842	0.025
	DCNet^［28］	0.808	0.730	0.805	0.851	0.025
	MPDC^［5］	0.817	0.751	0.812	0.837	0.018
	DECseg^［29］	0.843	0.792	0.839	0.863	0.010
	本文	0.877	0.820	0.872	0.901	0.013
Kvasir	URPC^［25］	0.739	0.644	0.727	0.783	0.067
	SSNet^［26］	0.769	0.670	0.750	0.807	0.064
	MCNet^［27］	0.779	0.681	0.758	0.808	0.063
	DCNet^［28］	0.790	0.695	0.768	0.813	0.061
	MPDC^［5］	0.801	0.712	0.785	0.834	0.052
	DECseg^［29］	0.863	0.797	0.858	0.883	0.039
	本文	0.877	0.805	0.870	0.888	0.038

Table 1

Fig. 4

Table 2

Comparison results on CVC-ColonDB and ETIS"

数据集	方法	评价指标
数据集	方法	mDice	mIoU	$F β ω$	$S α$	MAE
CVC-ColonDB	URPC^［25］	0.461	0.379	0.441	0.586	0.054
	SSNet^［26］	0.514	0.409	0.462	0.664	0.073
	MCNet^［27］	0.477	0.394	0.448	0.614	0.064
	DCNet^［28］	0.534	0.461	0.525	0.612	0.033
	MPDC^［5］	0.575	0.498	0.566	0.633	0.029
	DECseg^［29］	0.724	0.650	0.714	0.773	0.025
	本文	0.727	0.646	0.713	0.798	0.032
ETIS	URPC^［25］	0.353	0.290	0.337	0.483	0.029
	SSNet^［26］	0.406	0.317	0.358	0.598	0.040
	MCNet^［27］	0.488	0.404	0.450	0.624	0.032
	DCNet^［28］	0.419	0.358	0.406	0.509	0.022
	MPDC^［5］	0.433	0.364	0.420	0.534	0.021
	DECseg^［29］	0.570	0.500	0.544	0.640	0.016
	本文	0.674	0.592	0.642	0.781	0.017

Table 2

Table 3

Comparison results on ISIC 2018 dataset"

数据集	方法	评价指标
数据集	方法	mDice	mIoU	$F β ω$	$S α$	MAE
ISIC2018	URPC^［25］	0.847	0.761	0.820	0.826	0.085
	SSNet^［26］	0.865	0.783	0.837	0.840	0.076
	MCNet^［27］	0.855	0.771	0.831	0.834	0.081
	DCNet^［28］	0.853	0.774	0.842	0.836	0.083
	MPDC^［5］	0.847	0.758	0.817	0.827	0.081
	DECseg^［29］	0.867	0.786	0.825	0.844	0.072
	本文	0.874	0.793	0.832	0.848	0.071

Table 3

Fig. 5

Table 4

Comparison results on brain tumor dataset"

数据集	方法	评价指标
数据集	方法	mDice	mIoU	$F β ω$	$S α$	MAE
BrainMRI	URPC^［25］	0.694	0.591	0.682	0.782	0.011
	SSNet^［26］	0.698	0.590	0.687	0.793	0.012
	MCNet^［27］	0.706	0.606	0.701	0.784	0.010
	DCNet^［28］	0.712	0.616	0.705	0.785	0.010
	MPDC^［5］	0.676	0.586	0.671	0.742	0.011
	DECseg^［29］	0.811	0.724	0.822	0.859	0.006
	本文	0.814	0.720	0.823	0.871	0.007

Table 4

Fig. 6

Fig. 7

Table 5

References 30

[1]	李国威, 刘静, 曹慧, 等. 深度学习在结肠息肉图像分割中的研究综述[J]. 计算机科学与探索, 2025, 19(5):1198-1216.
	Li Guo-wei, Liu Jing, Cao Hui, et al. Research review of deep learning in colon polyp image segmentation[J]. Journal of Frontiers of Computer Science & Technology, 2025, 19(5):1198-1216.
[2]	Wang H L, Wang K N, Hua J, et al. Dynamic spectrum-driven hierarchical learning network for polyp segmentation[J]. Medical Image Analysis, 2025, 101: 103449.
[3]	Lv G Z, Wang B, Xu C L, et al. MFAINet: multi-receptive field feature fusion with attention-integrated for polyp segmentation[J]. IEEE/CAA Journal of Automatica Sinica, 2026: 13(4):822-836.
[4]	考文涛, 李明, 马金刚. 卷积神经网络在结直肠息肉辅助诊断中的应用综述[J]. 计算机科学与探索, 2024, 18(3):627-645.
	Wen-tao Kao, Li Ming, Ma Jin-gang. Review of application of convolutional neural network in auxiliary diagnosis of colorectal polyps[J]. Journal of Frontiers of Computer Science & Technology, 2024, 18(3):627-645
[5]	Zhu Y F, Wang X, Liu T H, et al. Multi-perspective dynamic consistency learning for semi-supervised medical image segmentation[J]. Scientific Reports, 2025, 15(1): 18266.
[6]	Assefa M, Naseer M, Ganapathi I I, et al. DyCON: dynamic uncertainty-aware consistency and contrastive learning for semi-supervised medical image segmentation[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2025: 30850-30860.
[7]	Sohn K, Berthelot D, Carlini N, et al. FixMatch: simplifying semi-supervised learning with consistency and confidence[C]∥Advances in Neural Information Processing Systems, Online, 2020: 596-608.
[8]	Tarvainen A, Valpola H. Mean teachers are better role models: weight-averaged consistency targets improve semi-supervised deep learning results[C]∥Advances in Neural Information Processing Systems, Long Beach, USA, 2017: 1195-1204.
[9]	Adame D, Nuñez J A, Vazquez F, et al. Topo-VM-UNetV2: encoding topology into vision mamba UNet for polyp segmentation[C]∥The 38th IEEE International Symposium on Computer-Based Medical Systems, Madrid, Spain, 2025: 258-263.
[10]	王雪, 李占山, 吕颖达. 基于多尺度感知和语义适配的医学图像分割算法[J]. 吉林大学学报:工学版, 2022, 52(3): 640-647.
	Wang Xue, Li Zhan-shan, Ying-da Lyu. Medical image segmentation based on multi⁃scale context⁃aware and semantic adaptor[J]. Journal of Jilin University(Engineering and Technology Edition), 2022, 52(3): 640-647.
[11]	才华, 王玉瑶, 付强, 等. 基于注意力机制和特征融合的语义分割网络[J]. 吉林大学学报:工学版, 2025, 55(4): 1384-1395.
	Cai Hua, Wang Yu-yao, Fu Qiang, et al. Semantic segmentation network based on attention mechanism and feature fusion[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(4):1384-1395.
[12]	Mu N, Zhang X C, Feng Y Z, et al. Towards robust polyp segmentation: multi-focus attention network with fine-grained polyp cues[C]∥Proceedings of the 2025 International Conference on Multimedia Retrieval, Chicago, USA,2025: 1027-1035.
[13]	Kamara A A, He S, Fofanah A J, et al. MDPNet: Multiscale dynamic polyp-focus network for enhancing medical image polyp segmentation[J]. IEEE Transactions on Medical Imaging, 2025,44(12): 5208-5220.
[14]	Zhou J Y, Yu W X, Zhang Z Q, et al. Context-aware multi-scale polyp segmentation network[C]∥IEEE International Conference on Acoustics Speech and Signal Processing, Hyderabad, India, 2025: 1-5.
[15]	Li Y X, Meng W Z, Ma D H, et al. MCGFF-Net: a multi-scale context-aware and global feature fusion network for enhanced polyp and skin lesion segmentation[J]. Visual Computer, 2025, 41(8): 5267-5282.
[16]	Wang W, Sun H Y, Wang X. LSSNet: A Method for colon polyp segmentation based on local feature supplementation and shallow feature supplementation[C]∥Medical Image Computing and Computer Assisted Intervention, Marrakesh, Morocco, 2024: 446-456.
[17]	Woo S, Park J, Lee J Y, et al. CBAM: convolutional block attention module[C]∥European Conference on Computer Vision, Munich, Germany, 2018: 3-19.
[18]	Chollet F. Xception: deep learning with depthwise separable convolutions[C]∥IEEE Conference on Computer Vision and Pattern Recognition,Honolulu, USA, 2017: 1800-1807.
[19]	Fan D P, Ji G P, Zhou T, et al. PraNet: parallel reverse attention network for polyp segmentation[C]∥Medical Image Computing and Computer Assisted Intervention, Lima, Peru, 2020: 263-273.
[20]	Tajbakhsh N, Gurudu S R, Liang J M. Automated polyp detection in colonoscopy videos using shape and context information[J]. IEEE Transactions on Medical Imaging, 2016, 35(2): 630-644.
[21]	Silva J, Histace A, Romain O, et al. Toward embedded detection of polyps in WCE images for early diagnosis of colorectal cancer[J]. International Journal of Computer Assisted Radiology and Surgery, 2014, 9(2): 283-293.
[22]	Codella N, Rotemberg V, Tschandl P,et al.Skin lesion analysis toward melanoma detection 2018: a challenge hosted by the international skin imaging collaboration(ISIC)[C]∥2018 IEEE 15th International Symposium on Biomedical Imaging,Washington DC, USA, 2018: 168-172.
[23]	Buda, Saha A, Mazurowski M A. Association of genomic subtypes of lower-grade gliomas with shape features automatically extracted by a deep learning algorithm[J]. Computers in Biology and Medicine, 2019, 109: 218-225.
[24]	Mazurowski M A, Clark K, Czarnek N M, et al. Radiogenomics of lower-grade glioma: algorithmically-assessed tumor shape is associated with tumor genomic subtypes and patient outcomes in a multi-institutional study with The Cancer Genome Atlas data[J]. Journal of Neuro-Oncology, 2017, 133(1): 27-35.
[25]	Luo X D, Liao W J, Chen J N, et al. Efficient semi-supervised gross target volume of nasopharyngeal carcinoma segmentation via uncertainty rectified pyramid consistency[C]∥Medical Image Computing and Computer Assisted Intervention, Strasbourg, France,2021: 318-329.
[26]	Wu Y C, Wu Z H, Wu Q Y, et al. Exploring smoothness and class-separation for semi-supervised medical image segmentation[C]∥Medical Image Computing and Computer Assisted Intervention,Singapore,2022: 34-43.
[27]	Wu Y C, Xu M F, Ge Z Y, et al. Semi-supervised left atrium segmentation with mutual consistency training[C]∥Medical Image Computing and Computer Assisted Intervention, Strasbourg, France, 2021: 297-306.
[28]	Chen F Q, Fei J J, Chen Y Q, et al. Decoupled consistency for semi-supervised medical image segmentation[C]∥Medical Image Computing and Computer Assisted Intervention, Vancouver, Canada, 2023: 551-561.
[29]	Gu Y Q, Zhou T, Zhang Y Z, et al. Dual-scale enhanced and cross-generative consistency learning for semi-supervised medical image segmentation[J]. Pattern Recognition, 2025, 158: 110962.
[30]	Gao S H, Cheng M M, Zhao K, et al. Res2Net: a new multi-scale backbone architecture[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(2): 652-662.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

Comments

Recommended 0

No Suggested Reading articles found!

评价指标	模块
评价指标	Baseline	+ECDM	+ECDM+CFEM	+ECDM+CFEM+MDFO
Parameters（M）	120.38	284.34	298.54	207.02
FLOPs（G）	90.89	174.57	199.26	159.17

Semi-supervised medical image segmentation algorithm based on dual-scale entangled complementary learning

RICH HTML

PDF (PC)

Abstract

Cite this article

share this article

Figures/Tables 12

References 30

Related Articles 4

Metrics

Comments

Recommended 0

[1]	Hong-wei ZHAO,Ming-zhu ZHOU,Ping-ping LIU,Qiu-zhan ZHOU. Medical image segmentation based on confident learning and collaborative training [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(5): 1675-1681.
[2]	Xiu-feng ZHANG,Yun-fei JIANG,Sheng-jin GUO,Yan-song LIU,Ling-zhuo TIAN,Shi-chen ZHANG. Brain tissue segmentation method combining multi-scale and attention mechanisms [J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(10): 3352-3360.
[3]	Xue WANG,Zhan-shan LI,Ying-da LYU. Medical image segmentation based on multi⁃scale context⁃aware and semantic adaptor [J]. Journal of Jilin University(Engineering and Technology Edition), 2022, 52(3): 640-647.
[4]	XIAO Ming-yao, LI Xiong-fei, ZHANG Xiao-li, ZHANG Liu. Medical image segmentation algorithm based on multi-scale region growing [J]. 吉林大学学报(工学版), 2017, 47(5): 1591-1597.