改进型密集递归残差U-Net的皮肤病变图像分割

doi:10.3969/j.issn.0258-8021.2025.03.004

Abstract
Figure/Table
References
Related Citation (1)

Download: PDF (8513 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract Accurate segmentation of skin lesion areas is essential for computer-aided diagnosis. However, irregular shapes, blurred boundaries and noise interference in skin lesion images lead to significant challenges to achieving high segmentation accuracy. To address these difficulties, we proposed an improved dense recurrent residual U-Net model (IDR2U-Net) for precise skin lesion area segmentation. Firstly, we optimized the original convolutional blocks in the encoding and decoding layers into recurrent residual convolution modules using dense connections to mitigate gradient vanishing problems in deep networks. Secondly, we introduced feature adaptation modules to strengthen the fusion of adjacent features by suppressing irrelevant background noise and amplifying informative features. Additionally, a dual attention mechanism was designed to enhance global information utilization efficiency using spatial attention and improve the correlation of channel features using channel attention, increasing network precision in skin lesion area segmentation. Furthermore, we mitigated class imbalance in dermoscopic image segmentation through joint Dice coefficient and cross-entropy loss function in our network training. Finally, ablation and comparative experiments were conducted using over 2 000 images from the ISIC 2017 skin lesion dataset. The IDR2U-Net model achieved Jaccard, Dice, and accuracy scores of 78.86%, 86.92%, and 94.61%, respectively. The experimental results demonstrated that the proposed IDR2U-Ne not only enhanced the accuracy but also achieved finer image segmentation, particularly in handling blurred boundary images, where it effectively reduced under-segmentation.

Key words： skin lesion image segmentation U-Net dense recurrent residual convolution module feature adaptation module dual attention mechanism

Received: 30 April 2024

PACS:

R318

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Zhao Dechun
	Yuan Yang
	Qin Lu
	Wei Li
	Ye Changrong

Cite this article:

Zhao Dechun,Yuan Yang,Qin Lu, et al. Improved Dense Recurrent Residual U-Netfor Image Segmentation of Skin Lesion[J]. Chinese Journal of Biomedical Engineering, 2025, 44(3): 291-300.

URL:

http://cjbme.csbme.org/EN/10.3969/j.issn.0258-8021.2025.03.004 OR http://cjbme.csbme.org/EN/Y2025/V44/I3/291

[1] Sung H, Ferlay J, Siegel R, 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.
[2] Kasmi R, Mokrani K. Classification of malignant melanoma and benign skin lesions: implementation of automatic ABCD rule[J]. IET Image Processing, 2016, 10 (6): 448-455.
[3] Dai Duwei, Dong Caixia, Xu Songhua, et al. Ms RED: a novel multi-scale residual encoding and decoding network for skin lesion segmentation[J]. Medical Image Analysis, 2022, 75: 102293.
[4] 尹稳, 周冬明, 范腾, 等. 基于密集空洞空间金字塔池化和注意力机制的皮肤病灶图像分割方法[J]. 生物医学工程学杂志, 2022, 39(6): 1108-1116.
[5] Liu Qi, Wang Jingkun, Zuo Mengying, et al. NCRNet: Neighborhood context refinement network for skin lesion segmentation[J]. Computers in Biology and Medicine, 2022, 146: 105545.
[6] Shan Pufang, Fu Chong, Dai Liming, et al. Automatic skin lesion classification using a new densely connected convolutional network with an SF module[J]. Medical & Biological Engineering & Computing, 2022, 60 (8): 2173-2188.
[7] Thanh DNH, Erkan U, Prasath S, et al. A skin lesion segmentation method for dermoscopic images based on adaptive thresholding with normalization of color models[C]//International Conference on Electrical and Electronic Engineering. Istanbul: IEEE, 2019: 116-120.
[8] Emin MY, Murat B. Accurate segmentation of dermoscopic images by image thresholding based on type-2 fuzzy logic[J]. IEEE Trans Fuzzy Systems, 2009, 17(4): 976-982.
[9] Alexander W, Jacob S, Paul F. Automatic skin lesion segmentation via iterative stochastic region merging[J]. IEEE Transactions on Information Technology in Biomedicine, 2011,15(6): 929-936.
[10] Jaisakthi, Seetharani M, Mirunalini, et al. Automated skin lesion segmentation of dermoscopic images using GrabCut and k-means algorithms[J]. IET Computer Vision, 2018, 12(8): 1088-1095.
[11] Ahn E, Kim J, Bi L, et al. Saliency-based lesion segmentation via background detection in dermoscopic images[J]. IEEE Journal of Biomedical and Health Informatics, 2017, 21(6): 1685-1693.
[12] 沈志强, 林超男, 潘林, 等. 基于同构化改进的U-Net结直肠息肉分割方法[J]. 中国生物医学工程学报, 2022, 41(1): 48-56.
[13] Xie Fengying, Yang Jiawen, Liu Jie, et al. Skin lesion segmentation using high-resolution convolutional neural network[J]. Computer Methods and Programs in Biomedicine, 2020, 186: 105241.
[14] 邓仕俊, 汤红忠, 曾黎, 等. 基于多尺度特征感知的胸腔图像危及器官分割[J]. 中国生物医学工程学报, 2021, 40(6): 701-711.
[15] Long J, Shelhamer E, Darrell T. Fully Convolutional Networks for Semantic Segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39(4): 640-651.
[16] Nasr-Esfahani E, Rafiei S, Mohammad H, et al. Dense pooling layers in fully convolutional network for skin lesion segmentation[J]. Computerized Medical Imaging and Graphics, 2019, 78:101658.
[17] Ronneberger O, Fischer P, Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation[C]//International Conference on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2015: 234-241.
[18] Ibtehaz N, Rahman MS. MultiResUNet: Rethinking the U-Net architecture for multimodal biomedical image segmentation[J]. Neural Networks, 2020, 121: 74-87.
[19] Rehman H, Nida N, Shah SA, et al. Automatic melanoma detection and segmentation in dermoscopy images using deep RetinaNet and conditional random fields[J]. Multimedia Tools and Applications, 2022, 81(18): 25765-25785.
[20] Gu Rui, Wang Lituan, Zhang Lei. DE-Net: a deep edge network with boundary information for automatic skin lesion segmentation[J]. Neurocomputing, 2022, 468: 71-84.
[21] Hu Kai, Lu Jing, Lee Dongjin, et al. AS-Net: attention synergy network for skin lesion segmentation[J]. Expert Systems with Applications, 2022, 201: 117112.
[22] Ling Wei, Fan Hong, Hu Chenxi, et al. Improved medical image segmentation model based on 3D U-Net[J]. Journal of Donghua University, 2022, 39(4): 311-316.
[23] Sarker MK, Rashwan HA, Akram F, et al. SLSNet: skin lesion segmentation using a lightweight generative adversarial network[J]. Expert Systems with Applications, 2021, 183: 115433.
[24] 张欢, 仇大伟, 冯毅博, 等. U-Net模型改进及其在医学图像分割上的研究综述[J]. 激光与光电子学进展, 2022, 59(2): 1-17.
[25] Alom Z, Asari VK, Parwani A, et al. Microscopic nuclei classification, segmentation, and detection with improved deep convolutional neural networks (DCNN)[J]. Diagnostic Pathology, 2022, 17(1): 1-17.
[26] Dutta K, Roy S, Whitehead TD, et al. Deep learning segmentation of triple-negative breast cancer (TNBC) patient derived tumor xenograft (PDX) and sensitivity of radiomic pipeline to tumor probability boundary[J]. Cancers, 2021, 13(15): 1-17.
[27] Wu Huisi, Chen Shihuai, Chen Guilian, et al. FAT-Net: feature adaptive transformers for automated skin lesion segmentation[J]. Medical Image Analysis, 2022, 76: 102327.
[28] Tran TT, Pham VT. Fully convolutional neural network with attention gate and fuzzy active contour model for skin lesion segmentation[J]. Multimedia Tools and Applications, 2022, 81(10): 13979-13999.
[29] 周涛, 董雅丽, 霍兵强, 等. U-Net网络医学图像分割应用综述[J]. 中国图象图形学报, 2021, 26(9): 2058-2077.
[30] Ma Shiqiang, Tang Jijun, Guo Fei. Multi-Task deep supervision on attention R2U-Net for brain tumor segmentation[J]. Frontiers in Oncology, 2021, 11: 704850.
[31] Wei Zenghui, Shi Feng, Song Hong, et al. Attentive boundary aware network for multi-scale skin lesion segmentation with adversarial training[J]. Multimedia Tools and Applications, 2020, 79(37-38): 27115-27136.
[32] Alom MZ, Yakopcic C, Taha, et al. Recurrent residual convolutional neural network based on U-Net (R2U-Net) for medical image segmentation[DB/OL]. http://arxiv.org/abs/1802.06955, 2018-02-20/2018-05-29.
[33] Vijay B, Alex K, Roberto C. SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495.
[34] Schlemper J, Oktay O, Schaap M, et al. Attention gated networks: learning to leverage salient regions in medical images[J]. Medical Image Analysis, 2019, 53: 197-207.
[35] Fu Jun, Liu Jing, Tian Haijie, et al. Dual attention network for scene segmentation[DB/OL]. http://arxiv.org/abs/1809.02983, 2019-04-21/2020-12-18.
[36] Xia Haiying, Wu Lingyu, Lan Yang, et al. HRNet: a hierarchical recurrent convolution neural network for retinal vessel segmentation[J]. Multimedia Tools and Applications, 2022, 81(28): 39829-39851.
[37] Zheng Shaohua, Nie Weiyu, Pan Lin, et al. A dual-attention V-network for pulmonary lobe segmentation in CT scans[J]. IET Image Processing, 2021, 15(8): 1644-1654.
[38] Li Wei, Raj ANJ, Tjahjadi T, et al. Digital hair removal by deep learning for skin lesion segmentation[J]. Pattern Recognition, 2021, 117: 107994.