[1]沈鑫,魏利胜.基于注意力残差U-Net的皮肤镜图像分割方法[J].智能系统学报,2023,18(4):699-707.[doi:10.11992/tis.202201030]
SHEN Xin,WEI Lisheng.Dermoscope image segmentation method based on ARB-UNet[J].CAAI Transactions on Intelligent Systems,2023,18(4):699-707.[doi:10.11992/tis.202201030]
点击复制
SHEN Xin,WEI Lisheng.Dermoscope image segmentation method based on ARB-UNet[J].CAAI Transactions on Intelligent Systems,2023,18(4):699-707.[doi:10.11992/tis.202201030]
基于注意力残差U-Net的皮肤镜图像分割方法
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
18
期数:
2023年第4期
页码:
699-707
栏目:
学术论文—机器学习
出版日期:
2023-07-15
- Title:
- Dermoscope image segmentation method based on ARB-UNet
- 关键词:
- 图像分割; 皮肤镜; 卷积神经网络; 注意力残差 U-Net; 注意力机制; 卷积块注意力机制模块; 深度学习; 残差网络
- Keywords:
- image segmentation; dermoscopic; convolutional neural network; attention residual block-UNet(ARB-UNet); attention mechanism; convolutional block attention module(CBAM); deep learning; residual network
- 分类号:
- TP391
- 摘要:
- 针对皮肤镜图像类内差异性、类间相似性、数据集不平衡等问题，本文提出了一种基于注意力残差U-Net（attention residual block-UNet，ARB-UNet）的皮肤镜图像分割方法。将卷积块注意力机制模块(convolutional block attention module,CBAM)引入到U-Net模型的“跳过连接”中；同时将CBAM模块集成到残差模块DRB（dilated residual networks）中得到注意力残差结构（attention residual block, ARB）；且选取Focal Tversky Loss作为该模型的损失函数；在ISIC2016数据集上对所提ARB-UNet模型进行训练和测试，并与传统方法和U-Net等经典方法进行了对比实验，实验结果中灵敏度（sensitivity，SE）达到了92.9%，特异性（specificity，SP）达到了94.1%，Dice相似指数（dice similarity cofficient，DSC）达到了92.1%，整体上均优于其他对比方法，从而验证了本文方法是有效的和可行的。
- Abstract:
- Aiming at the problems of intra-class difference, inter-class similarity, and dataset imbalance in dermoscopic images, a dermoscopic image segmentation method based on attention residual block-UNet (ARB-UNet) is proposed. Firstly, the convolutional block attention module (CBAM) is introduced into the “skip connection” of U-Net model; at the same time, the CBAM module is integrated into the residual module DRB (dilated residual networks) to obtain the attention residual block(ARB); Focal Tversky loss is selected as the loss function of the model; Finally, the proposed ARB-UNet model is trained and tested on ISIC2016 data set, and compared with traditional methods and classical methods such as U-Net. The experimental results show that the sensitivity (SE), specificity (SP), and dice similarity index (DSC) have reached 92.9%, 94.1%, and 92.1%, respectively, which are all better than other comparative methods in overall. Thus, the feasibility and effectiveness of the method in this paper are verified
- 参考文献/References:
-
[1] CI??Y?SKA M, KAMI?SKA-WINCIOREK G, LANGE D, et al. The incidence and clinical analysis of non-melanoma skin cancer[J]. Scientific reports, 2021, 11: 4337.
[2] LI Yuexiang, SHEN Linlin. Skin lesion analysis towards melanoma detection using deep learning network[J]. Sensors, 2018, 18(2): 556.
[3] SIEGEL R L, FEDEWA S A, MILLER K D, et al. Cancer statistics for hispanics/latinos, 2015[J]. CA:a cancer journal for clinicians, 2015, 65(6): 457–480.
[4] 徐国雄, 张骁, 胡进贤, 等. 基于阈值分割和轮廓提取的图像边缘检测算法[J]. 计算机技术与发展, 2015, 25(12): 64–67,71
XU Guoxiong, ZHANG Xiao, HU Jinxian, et al. Medical cell edge detection algorithm based on threshold segmentation and contour extraction[J]. Computer technology and development, 2015, 25(12): 64–67,71
[5] RIYANTO I. Morphology based melanoma skin segmentation method[J]. Helix, 2021, 11(3): 7–13.
[6] 尹雨山, 王李进, 尹义龙, 等. 回溯搜索优化算法辅助的多阈值图像分割[J]. 智能系统学报, 2015, 10(1): 68–74
YIN Yushan, WANG Lijin, YIN Yilong, et al. Backtracking search optimization algorithm assisted multilevel threshold for image segmentation[J]. CAAI transactions on intelligent systems, 2015, 10(1): 68–74
[7] 孔令军, 王茜雯, 包云超, 等. 基于深度学习的医疗图像分割综述[J]. 无线电通信技术, 2021, 47(2): 121–130
KONG Lingjun, WANG Qianwen, BAO Yunchao, et al. A survey on medical image segmentation based on deep learning[J]. Radio communciations technology, 2021, 47(2): 121–130
[8] STARVIN M S. A novel and efficient approach for the classification of skin melanoma[J]. Journal of ambient intelligence and humanized computing, 2021, 12(12): 10435–10459.
[9] LAKSHMI B V, SRIDEVI K. A qualitative approach on de-noising and segmentation algorithms for melanoma images[J]. Annals of the romanian society for cell biology, 2021: 1384-1393.
[10] FAN Haidi. Automatic segmentation of dermoscopy images using saliency combined with Otsu threshold[J]. Computers in biology and medicine, 2017, 85: 75–85.
[11] XIE Fengying, FAN Haidi, LI Yang, et al. Melanoma classification on dermoscopy images using a neural network ensemble model[J]. IEEE transactions on medical imaging, 2017, 36(3): 849–858.
[12] GLAISTER J, WONG A, CLAUSI D A. Segmentation of skin lesions from digital images using joint statistical texture distinctiveness[J]. IEEE transactions on bio-medical engineering, 2014, 61(4): 1220–1230.
[13] SHELHAMER E, LONG J, DARRELL T. Fully convolutional networks for semantic segmentation[C]//IEEE Transactions on Pattern Analysis and Machine Intelligence. Piscataway: IEEE, 2016: 640-651.
[14] RONNEBERGER O, FISCHER P, BROX T. U-net: convolutional networks for biomedical image segmentation[M]. Cham: Springer International Publishing, 2015: 234?241.
[15] ZHOU Zongwei, RAHMAN SIDDIQUEE M M, TAJBAKHSH N, et al. UNet++: A nested U-net architecture for medical image segmentation[M]. Cham: Springer International Publishing, 2018: 3?11.
[16] OKTAY O, SCHLEMPER J, FOLGOC L L, et al. Attention U-net: learning where to look for the pancreas[EB/OL]. (2018-04-11)[2022-01-18].https://arxiv.org/abs/1804.03999.
[17] SARKER M M K, RASHWAN H A, AKRAM F, et al. SLSDeep: skin lesion segmentation based on dilated residual and pyramid pooling networks[M]. Cham: Springer International Publishing, 2018: 21?29.
[18] ZHANG Lei, YANG Guang, YE Xujiong. Automatic skin lesion segmentation by coupling deep fully convolutional networks and shallow network with textons[J]. Journal of medical imaging, 2019, 6(2): 024001.
[19] BAGHERSALIMI S, BOZORGTABAR B, SCHMID-SAUGEON P, et al. DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation[J]. EURASIP journal on image and video processing, 2019, 2019(1): 71.
[20] SELLA VELUSWAMI J R, EZHIL PRASANTH M, HARINI K, et al. Melanoma skin cancer recognition and classification using deep hybrid learning[J]. Journal of medical imaging and health informatics, 2021, 11(12): 3110–3116.
[21] LIU Jing, CHEN Aibin, ZHOU Guoxiong, et al. Dermatoscopic image melanoma recognition based on CFLDnet fusion network[J]. Multimedia tools and applications, 2021, 80(17): 25477–25494.
[22] GUO Changlu, SZEMENYEI M, YI Yugen, et al. Dense residual network for retinal vessel segmentation[C]//2020 IEEE International Conference on Acoustics, Speech and Signal Processing. Piscataway: IEEE, 2020: 1374-1378.
[23] KAUR R, GHOLAMHOSSEINI H, SINHA R, et al. Melanoma classification using a novel deep convolutional neural network with dermoscopic images[J]. Sensors (Basel, Switzerland), 2022, 22(3): 1134.
[24] WANG Hailing, YANG Jingdong. FBUNet: full convolutional network based on fusion block architecture for biomedical image segmentation[J]. Journal of medical and biological engineering, 2021, 41(2): 185–202.
[25] CHEN Ziming, WANG Shengsheng. The NL-SC net for skin lesion segmentation[M].Pattern Recognition and Computer Vision. Cham: Springer International Publishing, 2021: 313?324.
- 相似文献/References:
-
[1]王科俊,郭庆昌.基于粒子群优化算法和改进的Snake模型的图像分割算法[J].智能系统学报,2007,2(1):53.
WANG Ke-jun,GUO Qing-chang.Image segmentation algorithm based on the PSO and improved Snake model[J].CAAI Transactions on Intelligent Systems,2007,2():53.
[2]陈小波,程显毅.一种基于MAS的自适应图像分割方法[J].智能系统学报,2007,2(4):80.
CHEN Xiao-bo,CHENG Xian-yi.An adaptive image segmentation technique based on multiAgent system[J].CAAI Transactions on Intelligent Systems,2007,2():80.
[3]刘咏梅,代丽洁.基于空间位置约束的K均值图像分割[J].智能系统学报,2010,5(1):67.
LIU Yong-mei,DAI Li-jie.An improved method of Kmeans image segmentation based on spatial position information[J].CAAI Transactions on Intelligent Systems,2010,5():67.
[4]吴一全,纪守新.灰度熵和混沌粒子群的图像多阈值选取[J].智能系统学报,2010,5(6):522.
WU Yi-quan,JI Shou-xin.Multithreshold selection for an image based on gray entropy and chaotic particle swarm optimization[J].CAAI Transactions on Intelligent Systems,2010,5():522.
[5]尚倩,阮秋琦,李小利.双目立体视觉的目标识别与定位[J].智能系统学报,2011,6(4):303.
SHANG Qian,RUAN Qiuqi,LI Xiaoli.Target recognition and location based on binocular stereo vision[J].CAAI Transactions on Intelligent Systems,2011,6():303.
[6]胡光龙,秦世引.动态成像条件下基于SURF和Mean shift的运动目标高精度检测[J].智能系统学报,2012,7(1):61.
HU Guanglong,QIN Shiyin.High precision detection of a mobile object under dynamic imaging based on SURF and Mean shift[J].CAAI Transactions on Intelligent Systems,2012,7():61.
[7]马慧,王科俊.采用旋转校正的指静脉图像感兴趣区域提取方法[J].智能系统学报,2012,7(3):230.
MA Hui,WANG Kejun.A region of interest extraction method using rotation rectified finger vein images[J].CAAI Transactions on Intelligent Systems,2012,7():230.
[8]尹雨山,王李进,尹义龙,等.回溯搜索优化算法辅助的多阈值图像分割[J].智能系统学报,2015,10(1):68.[doi:10.3969/j.issn.1673-4785.201410008]
YIN Yushan,WANG Lijin,YIN Yilong,et al.Backtracking search optimization algorithm assisted multilevel threshold for image segmentation[J].CAAI Transactions on Intelligent Systems,2015,10():68.[doi:10.3969/j.issn.1673-4785.201410008]
[9]吴一全,王凯,曹鹏祥.蜂群优化的二维非对称Tsallis交叉熵图像阈值选取[J].智能系统学报,2015,10(1):103.[doi:10.3969/j.issn.1673-4785.201403040]
WU Yiquan,WANG Kai,CAO Pengxiang.Two-dimensional asymmetric tsallis cross entropy image threshold selection using bee colony optimization[J].CAAI Transactions on Intelligent Systems,2015,10():103.[doi:10.3969/j.issn.1673-4785.201403040]
[10]龙鹏,鲁华祥.方差不对称先验信息引导的全局阈值分割方法[J].智能系统学报,2015,10(5):663.[doi:10.11992/tis.201412022]
LONG Peng,LU Huaxiang.Global threshold segmentation technique guided by prior knowledge with asymmetric variance[J].CAAI Transactions on Intelligent Systems,2015,10():663.[doi:10.11992/tis.201412022]
备注/Memo
收稿日期:2022-01-18。
基金项目:安徽省教育厅重大项目(KJ2020ZD39）；安徽省检测技术与节能装置重点实验室开放基金项目(DTESD2020A02).
作者简介:沈鑫，硕士研究生，主要研究方向为智能信息处理;魏利胜，教授，博士，主要研究方向为图像识别与应用、嵌入式仪器仪表及系统、智能化网络控制理论、系统和仿真。主持国家自然科学基金项目、安徽省自然科学基金项目、安徽省教育厅重大项目等省部级以上项目8项，授权发明专利6项，获得安徽省科技进步一等奖1项。发表学术论文100余篇。
通讯作者:魏利胜.E-mail:lshwei_11@163.com
更新日期/Last Update:
1900-01-01