吉林大学学报(工学版) ›› 2023, Vol. 53 ›› Issue (12): 3565-3572.doi: 10.13229/j.cnki.jdxbgxb.20220141

• 计算机科学与技术 • 上一篇    

基于双输入3D卷积神经网络的胰腺分割算法

刘桂霞1,2(),田郁欣1,2,王涛1,2,马明睿1,2   

  1. 1.吉林大学 计算机科学与技术学院,长春 130012
    2.吉林大学 符号计算与知识工程教育部重点实验室,长春 130012
  • 收稿日期:2022-02-16 出版日期:2023-12-01 发布日期:2024-01-12
  • 作者简介:刘桂霞(1963-),女,教授,博士.研究方向:机器学习和医学图像分析.E-mail:liugx@jlu.edu.cn
  • 基金资助:
    国家自然科学基金项目(61772226);吉林省科技厅重点项目(20210204133YY);吉林省自然科学基金项目(20200201159JC)

Pancreas segmentation algorithm based on dual input 3D convolutional neural network

Gui-xia LIU1,2(),Yu-xin TIAN1,2,Tao WANG1,2,Ming-rui MA1,2   

  1. 1.College of Computer Science and Technology,Jilin University,Changchun 130012,China
    2.Key Laboratory of Symbol Computation and Knowledge Engineer of Ministry of Education,Jilin University,Changchun 130012,China
  • Received:2022-02-16 Online:2023-12-01 Published:2024-01-12

摘要:

针对胰腺分割难以获得较高准确率的问题,本文提出了一种基于双输入3D卷积神经网络的胰腺分割算法。首先,通过增加输入切片间上下文残差信息突出边界区域;然后,引入注意力机制抑制无用特征、增加有效特征的表达,提高了胰腺的分割准确率;最后,将该算法在NIH胰腺分割数据集上进行评估。实验结果表明,本文算法性能优先于对比的主流算法。

关键词: 计算机应用, 医学图像, 胰腺分割, 上下文残差信息, 注意力机制

Abstract:

Addressing the issue of difficulty in achieving high accuracy in achieving high accuracy in pancreas segmentation, a dual input 3D network to realize panc-reatic segmentation was proposed. Firstly, the boundary area was highlighted by increasing the context residual information between input slices, and then the attention mechanism was introduced to suppress useless features and increase the expression of effective features, which finally improve the accuracy of pancreatic segmentation. The algorithm was evaluated on NIH pancreas segmentation data set. The experiment-al results show that the performance of the algorithm proposed in this paper is superior to the comparison algorithm.

Key words: computer application, medical image, pancreas segmentation, context residual information, attention mechanism

中图分类号: 

  • TP391

图1

本文算法网络结构"

图2

残差模块"

图3

挤压和激励模块"

图4

上下文残差信息"

图5

本文算法分割结果"

表1

不同条件下消融实验结果"

实验序号架构包含模块Dice系数 平均值/%
1残差模块82.84±8.98
2残差模块+SE模块83.39±6.74
3残差模块+上下文残差信息83.57±5.88
4残差模块+上下文残差信息+SE模块84.37±5.14

图6

不同实验的可视化对比"

表2

不同轴向的上下文残差信息性能对比"

网络包含模块视图平均Dice系数/%最大Dice系数/%最小Dice系数/%
残差模块+SE模块-83.39±6.7490.8853.38
残差模块+SE模块+上下文残差信息水平面84.37±5.1491.6556.35
残差模块+SE模块+上下文残差信息矢状面84.29±5.5791.2154.95
残差模块+SE模块+上下文残差信息冠状面84.47±5.1492.0750.86

图7

不同轴向的可视化对比"

表3

不同胰腺分割方法性能对比"

方法Mean Dice/%Max Dice/%Min Dice/%K-CV
文献[881.27±6.2788.9650.694-CV
文献[981.48±6.23--5-CV
文献[783.18±4.8191.0365.104-CV
文献[1183.70±5.1091.0059.004-CV
文献[1284.10±4.91--4-CV
本文84.37±5.1491.6556.354-CV
1 Zhang Y, Wu J, Liu Y, et al. A deep learning framework for pancreas segmentation with multi-atlas registration and 3D level-set[J]. Medical Image Analysis, 2021, 68: No. 101884.
2 Pandey S, Tekchandani H, Verma S. A literature review on application of machine learning techniques in pancreas segmentation[C]∥2020 First International Conference on Power, Control and Computing Technologies (ICPC2T), Raipur, India, 2020: 401-405.
3 Bazgir O, Barck K, Carano R A D, et al. Kidney segmentation using 3D U-Net localized with Expectation Maximization[C]∥2020 IEEE Southwest Symposium on Image Analysis and Interpretation (SSIAI), Albuquerque, NM, USA, 2020: 22-25.
4 Ibrahim S M, Ibrahim M S, Usman M, et al. A study on heart segmentation using deep learning algorithm for mri scans[C]∥2019 13th International Conference on Mathematics, Actuarial Science, Computer Science and Statistics (MACS), Normandy, France, 2019: 1-5.
5 Li C, Tan Y, Chen W, et al. Attention Unet++: a nested attention-aware U-net for liver CT image segmentation[C]∥2020 IEEE International Conference on Image Processing (ICIP), Abu Dhabi, United Arab Emirates, 2020: 345-349.
6 Man Y, Huang Y, Feng J, et al. Deep Q learning driven CT pancreas segmentation with geometry-aware U-Net[J]. IEEE Transactions on Medical Imaging, 2019, 38(8): 1971-1980.
7 Zhou Y, Xie L, Shen W, et al. A fixed-point model for pancreas segmentation in abdominal CT scans[C]∥International Conference on Medical Image Computing and Computer-Assisted Intervention, Quebec, Canada, 2017: 693-701.
8 Roth H R, Lu L, Lay N, et al. Spatial aggregation of holistically-nested convolutional neural networks for automated pancreas localization and segmentation[J]. Medical Image Analysis, 2018, 45: 94-107.
9 Oktay O, Schlemper J, Folgoc L L, et al. Attention U-Net: learning where to look for the pancreas[J/OL]. [2022-01-25].
10 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, Munich, Germany, 2015: 234-241.
11 Cai J, Lu L, Xing F, et al. Pancreas segmentation in CT and MRI images via domain specific network designing and recurrent neural contextual learning[J/OL]. [2022-01-28].
12 Liu S, Yuan X, Hu R, et al. Automatic pancreas segmentation via coarse location and ensemble learning[J]. IEEE Access, 2019, 8: 2906-2914.
13 Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Boston, USA, 2015: 3431-3440.
14 凌语, 孙自强. 基于卷积神经网络的乳腺病理图像识别算法[J]. 江苏大学学报: 自然科学版, 2019, 40(5): 573-578.
Ling Yu, Sun Zi-qiang. Breast pathological image recognition algorithm based on convolutional neural network[J]. Journal of Jiangsu University (Natural Science Edition), 2019, 40(5): 573-578.
15 He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 770-778.
16 He K, Zhang X, Ren S, et al. Identity mappings in deep residual networks[C]∥European Conference on Computer Vision, Amsterdam, Netherlands, 2016: 630-645.
17 Wu Y, He K. Group normalization[C]∥Proceedings of the European conference on computer vision (ECCV), Munich, Germany, 2018: 3-19.
18 Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, Utah, 2018: 7132-7141.
19 陈洁, 詹永照. 多时间尺度双流CNN与置信融合的视频动作识别[J]. 江苏大学学报: 自然科学版, 2021, 42(3): 318-324.
Chen Jie, Zhan Yong-zhao. Video motion recognition fused with multi-time-scale dual-stream CNN and confidence[J]. Journal of Jiangsu University (Natural Science Edition), 2021, 42(3): 318-324.
20 Howard A, Sandler M, Chu G, et al. Searching for mobilenetv3[C]∥Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, Korea, 2019: 1314-1324.
21 Yu F, Koltun V, Funkhouser T. Dilated residual networks[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Hawaii, 2017: 472-480.
22 Zhang J, Xie Y, Wang Y, et al. Inter-slice context residual learning for 3D medical image segmentation[J]. IEEE Transactions on Medical Imaging, 2020, 40(2): 661-672.
23 Roth H R, Lu L, Farag A, et al. Deep organ: Multi-level deep convolutional networks for automated pancreas segmentation[C]∥International Conference on Medical Image Computing and Computer-assisted Intervention, Munich, Germany, 2015: 556-564.
[1] 霍光,林大为,刘元宁,朱晓冬,袁梦,盖迪. 基于多尺度特征和注意力机制的轻量级虹膜分割模型[J]. 吉林大学学报(工学版), 2023, 53(9): 2591-2600.
[2] 何颖,王卓然,周旭,刘衍珩. 融合社交地理信息加权矩阵分解的兴趣点推荐算法[J]. 吉林大学学报(工学版), 2023, 53(9): 2632-2639.
[3] 张云佐,董旭,蔡昭权. 拟合下肢几何特征的多视角步态周期检测[J]. 吉林大学学报(工学版), 2023, 53(9): 2611-2619.
[4] 肖明尧,李雄飞,朱芮. 基于NSST域像素相关分析的医学图像融合[J]. 吉林大学学报(工学版), 2023, 53(9): 2640-2648.
[5] 赵亚慧,李飞雨,崔荣一,金国哲,张振国,李德,金小峰. 基于跨语言预训练模型的朝汉翻译质量评估[J]. 吉林大学学报(工学版), 2023, 53(8): 2371-2379.
[6] 郭晓新,李佳慧,张宝亮. 基于高分辨率网络的视杯和视盘的联合分割[J]. 吉林大学学报(工学版), 2023, 53(8): 2350-2357.
[7] 车翔玖,徐欢,潘明阳,刘全乐. 生物医学命名实体识别的两阶段学习算法[J]. 吉林大学学报(工学版), 2023, 53(8): 2380-2387.
[8] 王连明,吴鑫. 基于姿态估计的物体3D运动参数测量方法[J]. 吉林大学学报(工学版), 2023, 53(7): 2099-2108.
[9] 唐菲菲,周海莲,唐天俊,朱洪洲,温永. 融合动静态变量的滑坡多步位移预测方法[J]. 吉林大学学报(工学版), 2023, 53(6): 1833-1841.
[10] 张振海,季坤,党建武. 基于桥梁裂缝识别模型的桥梁裂缝病害识别方法[J]. 吉林大学学报(工学版), 2023, 53(5): 1418-1426.
[11] 刘培勇,董洁,谢罗峰,朱杨洋,殷国富. 基于多支路卷积神经网络的磁瓦表面缺陷检测算法[J]. 吉林大学学报(工学版), 2023, 53(5): 1449-1457.
[12] 张则强,梁巍,谢梦柯,郑红斌. 混流双边拆卸线平衡问题的精英差分进化算法[J]. 吉林大学学报(工学版), 2023, 53(5): 1297-1304.
[13] 田彦涛,黄兴,卢辉遒,王凯歌,许富强. 基于注意力与深度交互的周车多模态行为轨迹预测[J]. 吉林大学学报(工学版), 2023, 53(5): 1474-1480.
[14] 于鹏,朴燕. 基于多尺度特征的行人重识别属性提取新方法[J]. 吉林大学学报(工学版), 2023, 53(4): 1155-1162.
[15] 姜宇,潘家铮,陈何淮,符凌智,齐红. 基于分割方法的繁体中文报纸文本检测[J]. 吉林大学学报(工学版), 2023, 53(4): 1146-1154.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!