吉林大学学报(工学版) ›› 2021, Vol. 51 ›› Issue (4): 1414-1419.doi: 10.13229/j.cnki.jdxbgxb20200204

• 计算机科学与技术 • 上一篇    

基于条件生成对抗网络的医学细胞图像生成检测方法

陈雪云(),许韬,黄小巧   

  1. 广西大学 电气工程学院,南宁 530004
  • 收稿日期:2020-04-01 出版日期:2021-07-01 发布日期:2021-07-14
  • 作者简介:陈雪云(1969-),男,副教授,博士.研究方向:机器学习与模式识别.E-mail:cxyn2020@163.com
  • 基金资助:
    国家自然科学基金项目(61661006)

Detection method of medical cell image generation based on conditional generative adversarial network

Xue-yun CHEN(),Tao XU,Xiao-qiao HUANG   

  1. School of Electrical Engineering,Guangxi University,Nanning 530004,China
  • Received:2020-04-01 Online:2021-07-01 Published:2021-07-14

摘要:

针对现有医学细胞图像生成检测方法在检测中需要大量的有标签数据作为支撑,而细胞在黏附遮挡的情况下数据匮乏,不利于细胞检测精度的提高这一问题,提出了基于条件生成对抗网络的细胞图像生成检测方法。通过Pix2pix网络模型控制生成黏附遮挡的细胞图像,提取损失函数,采用Pix2pix实现图像到图像的转换,并运用正则项误差控制生成对抗网络误差。在此基础上,构建检测网络,包括生成网络结构、判别网络结构和检测网络结构,在生成网络输出端进行目标检测,使图像生成与细胞检测工作在同一个网络中完成。实验表明,与现有模型相比,本文方法在检测精度上有显著提升,达到了90.2%,可以满足医学细胞检测需求。

关键词: 条件生成对抗网络, 图像生成, 目标检测, 判别网络结构, 损失函数

Abstract:

The existing methods need a lot of labeled data as support in the detection, but the lack of data in the case of cell adhesion and occlusion is not conducive to the improvement of cell detection accuracy. In order to solve this problem, a cell image generation detection method based on condition generation antagonism network is proposed. Pix2pix network model is used to control the generation of cell image with adhesion occlusion, the loss function is extracted, pix2pix is used to realize image to image conversion, and regular term error control is used to generate network error. On this basis, the detection network is constructed, including the generation network structure, discrimination network structure and detection network structure. The target detection is carried out at the output of the generation network, so that the image generation and cell detection are completed in the same network. Experiments show that compared with the existing model, the designed method has a significant improvement in detection accuracy, reaching 90.2%, which can meet the needs of medical cell detection.

Key words: conditional generative adversarial network, image generation, target detection, discriminant network structure, loss function

中图分类号: 

  • TP391.4

图1

LCGAN流程图"

图2

网络框架图"

图3

判别网络结构"

图4

不同模型生成细胞数据库"

表1

PSNR、SSIM、IS定量结果对比"

方法名称迭代次数PSRNSSIMIS
Pix2pix50 00017.540.8430.79
LCGAN50 00020.870.9050.82

图5

不同模型检测细胞数据库"

表2

各检测方法实验对比结果"

方法名称迭代次数召回率/%精确率/%准确率
BP+HOG10 00068.485.80.77
SSD10 00078.389.20.83
FCRN10 00090.391.20.894
LCGAN10 00091.291.70.902
1 万程,周鹏,吴陆辉,等.基于生成对抗网络的糖尿病视网膜病变眼底图像生成[J].中华实验眼科杂志,2019,37(8):613-618.
Wan Cheng, Zhou Peng, Wu Lu-hui, et al. Diabetic retinopathy fundus image generation based on generative adversarial networks[J]. Chinese Journal of Experimental Ophthalmology, 2019,37(8):613-618.
2 胡海洋,庄载雄,俞佳成,等.基于生成对抗网络的模糊工件角度检测[J].计算机集成制造系统,2019,25(8):1936-1945.
Hu Hai-yang, Zhuang Zai-xiong, Yu Jia-cheng, et al. Blurred workpiece angle detection method based on generative adversarial networks[J]. Computer Integrated Manufacturing Systems, 2019,25(8):1936-1945.
3 Imani Maryam. Hyperspectral anomaly detection using differential image[J]. IET Image Processing, 2018, 12(5):801-809.
4 苏健民,杨岚心.基于生成对抗网络的单帧遥感图像超分辨率[J].计算机工程与应用,2019,55(12):202-207, 214.
Su Jian-min, Yang Lan-xin. Super resolution of single frame remote sensing image based on generative adversarial nets[J]. Computer Engineering and Applications, 2019,55(12):202-207, 214.
5 王海鸥,刘慧,郭强,等.面向医学图像分割的超像素U-Net网络设计[J].计算机辅助设计与图形学学报,2019,31(6):1007-1017.
Wang Hai-ou, Liu Hui, Guo Qiang, et al. Design of superpiexl U-Net network for medical image segmentation[J]. Journal of Computer-Aided Design and Computer Graphics, 2019,31(6):1007-1017.
6 Goodfellow I,Pouget-Abadie J, Mirza M, et al. Generative adversarial nets[C]∥Advances in Neural Information Processing Systems, Barcelona Spain, 2014: 2672-2680
7 Radford A, Metz L, Chintala S. Unsupervised representation learning with deep convolutional generative adversarial networks[J]. arXiv preprint arXiv:1511.06434, 2015
8 Mirza M, Osindero S. Conditional generative adversarial nets[J]. Computer Science, 2014:2672-2680.
9 肖进胜,申梦瑶,雷俊锋,等.基于生成对抗网络的雾霾场景图像转换算法[J].计算机学报,2020,43(1):165-176.
Xiao Jin-sheng, Shen Meng-yao, Lei Jun-feng, et al. Image conversion algorithm for haze scene based on generative adversarial networks[J]. Chinese Journal of Computers, 2020,43(1):165-176
10 Zhao H, Gallo O, Frosio I, et al. Loss functions for image restoration with neural networks[J]. Computational Imaging, IEEE Transactions on, 2017, 3(1):47-57.
11 霍其润,李建武,陆耀,等.基于变分的CT图像环形伪影校正[J].自动化学报,2019,45(9):1713-1726.
Huo Qi-run, Li Jian-wu, Lu Yao, et al. Variation-based ring artifact correction in CT images[J]. Acta Automatica Sinica, 2019,45(9):1713-1726.
12 于剑楠,李辉勇,李青锋,等.DM+:一种融合数字编码的可扫描图像生成技术[J].计算机学报,2019,42(9):2035-2048.
Yu Jian-nan, Li Hui-yong, Li Qing-feng, et al. DM+: a scanning image generation technique based on digital coding[J]. Chinese Journal of Computers, 2019,42(9):2035-2048.
13 张凯,李军辉,周国栋.基于枢轴语言的图像描述生成研究[J].中文信息学报,2019,33(3):110-117.
Zhang Kai, Li Jun-hui, Zhou Guo-dong. Image caption via pivot language[J]. Journal of Chinese Information Processing, 2019,33(3):110-117.
14 刘如飞,朱健,杨正清,等.一种针对路面损坏识别的点云特征图像生成方法[J].遥感信息,2019,34(4):22-28.
Liu Ru-fei, Zhu Jian, Yang Zheng-qing, et al. A point cloud feature image generation method for pavement damage recognition[J]. Remote Sensing Information, 2019,34(4):22-28.
[1] 潘德伦,冀隽,张跃进. 基于运动矢量空间编码的视频监控动态目标检测方法[J]. 吉林大学学报(工学版), 2021, 51(4): 1370-1374.
[2] 刘元宁,吴迪,朱晓冬,张齐贤,李双双,郭书君,王超. 基于YOLOv3改进的用户界面组件检测算法[J]. 吉林大学学报(工学版), 2021, 51(3): 1026-1033.
[3] 袁哲明,袁鸿杰,言雨璇,李钎,刘双清,谭泗桥. 基于深度学习的轻量化田间昆虫识别及分类模型[J]. 吉林大学学报(工学版), 2021, 51(3): 1131-1139.
[4] 徐涛,马克,刘才华. 基于深度学习的行人多目标跟踪方法[J]. 吉林大学学报(工学版), 2021, 51(1): 27-38.
[5] 赵宏伟,王鹏,范丽丽,胡黄水,刘萍萍. 相似性保持实例检索方法[J]. 吉林大学学报(工学版), 2019, 49(6): 2045-2050.
[6] 车翔玖,刘华罗,邵庆彬. 基于Fast RCNN改进的布匹瑕疵识别算法[J]. 吉林大学学报(工学版), 2019, 49(6): 2038-2044.
[7] 徐谦,李颖,王刚. 基于深度学习的行人和车辆检测[J]. 吉林大学学报(工学版), 2019, 49(5): 1661-1667.
[8] 黄勇,杨德运,乔赛,慕振国. 高分辨合成孔径雷达图像的耦合传统恒虚警目标检测[J]. 吉林大学学报(工学版), 2018, 48(6): 1904-1909.
[9] 杨超宇, 李策, 梁胤程, 杨峰. 基于改进粒子滤波的煤矿视频监控模糊目标检测[J]. 吉林大学学报(工学版), 2017, 47(6): 1976-1985.
[10] 姜宏, 李垠, 吕巍. 基于线性收缩的大阵列MIMO雷达目标盲检测[J]. 吉林大学学报(工学版), 2017, 47(3): 973-980.
[11] 孙挺, 齐迎春, 耿国华. 基于帧间差分和背景差分的运动目标检测算法[J]. 吉林大学学报(工学版), 2016, 46(4): 1325-1329.
[12] 陈大伟, 刘海龙, 李金屏. 复杂静态背景下多移动目标实时检测系统的FPGA实现[J]. 吉林大学学报(工学版), 2013, 43(增刊1): 287-290.
[13] 史东承, 闫李. 基于场景知识的移动目标检测[J]. 吉林大学学报(工学版), 2013, 43(增刊1): 275-278.
[14] 秦传东, 刘三阳. 基于双重正则化支持向量机的肿瘤基因选择[J]. 吉林大学学报(工学版), 2013, 43(01): 192-197.
[15] 孙瑾秋, 张艳宁, 周军, 姜磊. 基于选择性视觉注意机制和曲线发现法的弱小目标检测[J]. 吉林大学学报(工学版), 2010, 40(06): 1683-1687.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!