《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (1): 302-309.DOI: 10.11772/j.issn.1001-9081.2021020258
收稿日期:
2021-02-22
修回日期:
2021-04-28
接受日期:
2021-04-29
发布日期:
2021-05-12
出版日期:
2022-01-10
通讯作者:
邓爽
作者简介:
邓爽 (1995—),女,四川绵阳人,硕士研究生,主要研究方向:图像处理、模式识别、人工智能基金资助:
Shuang DENG(), Xiaohai HE, Linbo QING, Honggang CHEN, Qizhi TENG
Received:
2021-02-22
Revised:
2021-04-28
Accepted:
2021-04-29
Online:
2021-05-12
Published:
2022-01-10
Contact:
Shuang DENG
About author:
DENG Shuang, born in 1995, M. S. candidate. Her research interests include image processing, pattern recognition, artificial intelligence.Supported by:
摘要:
针对阿尔兹海默症(AD)患者和正常(NC)人之间核磁共振成像(MRI)图像差别小、分类难度大的问题,提出了基于改进VGG网络的弱监督细粒度AD分类方法。该方法以弱监督数据增强网络(WSDAN)为基本模型,主要由弱监督注意力学习模块、数据增强模块及双线性注意力池化模块等构成。首先,通过弱监督力注意学习模块生成特征图和注意力图,并利用注意力图引导数据增强,将原图和增强后的数据同时作为输入数据进行训练;然后,通过双线性注意力池化算法将特征图和注意力图按元素进行点乘,进而得到特征矩阵;最后,将特征矩阵作为线性分类层的输入。将以VGG19作为特征提取网络的WSDAN基本模型应用到AD的MRI数据上,实验结果表明,仅使用图像增强的模型的准确性、敏感性和特异性分别比WSDAN基本模型提高了1.6个百分点、0.34个百分点和0.12个百分点;仅利用VGG19网络的改进的模型的准确性和特异性相较WSDAN基本模型分别提高了0.7个百分点和2.82个百分点;以上两个方法结合使用的模型与WSDAN基本模型相比,准确性、敏感性和特异性分别提高了2.1个百分点、1.91个百分点和2.19个百分点。
中图分类号:
邓爽, 何小海, 卿粼波, 陈洪刚, 滕奇志. 基于改进VGG网络的弱监督细粒度阿尔兹海默症分类方法[J]. 计算机应用, 2022, 42(1): 302-309.
Shuang DENG, Xiaohai HE, Linbo QING, Honggang CHEN, Qizhi TENG. Weakly supervised fine-grained classification method of Alzheimer’s disease based on improved visual geometry group network[J]. Journal of Computer Applications, 2022, 42(1): 302-309.
卷积层 | 通道数 | 网络参数 |
---|---|---|
Conv1 | 64 | kernel: |
MaxPool1 | 64 | kernel: |
Conv2 | 128 | kernel: |
MaxPool2 | 128 | kernel: |
Conv3 | 256 | kernel: |
MaxPool3 | 256 | kernel: |
Conv4 | 512 | kernel: |
MaxPool4 | 512 | kernel: |
Conv5 | 512 | kernel: |
MaxPool5 | 512 | kernel: |
表1 VGG19网络参数
Tab.1 VGG19 network parameters
卷积层 | 通道数 | 网络参数 |
---|---|---|
Conv1 | 64 | kernel: |
MaxPool1 | 64 | kernel: |
Conv2 | 128 | kernel: |
MaxPool2 | 128 | kernel: |
Conv3 | 256 | kernel: |
MaxPool3 | 256 | kernel: |
Conv4 | 512 | kernel: |
MaxPool4 | 512 | kernel: |
Conv5 | 512 | kernel: |
MaxPool5 | 512 | kernel: |
模型 | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
VGG19 | 92.20 | 92.49 | 91.87 |
ResNet101 | 91.90 | 92.81 | 90.93 |
表2 传统的分类网络性能对比 (%)
Tab.2 Performance comparison of traditional classification networks
模型 | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
VGG19 | 92.20 | 92.49 | 91.87 |
ResNet101 | 91.90 | 92.81 | 90.93 |
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN(ResNet101) | 95.10 | 97.40 | 92.81 |
WSDAN (Inception) | 94.80 | 95.60 | 94.06 |
表3 使用不同特征提取网络的WSDAN基础网络模型 (%)
Tab.3 WSDAN basic network models with different feature extraction networks
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN(ResNet101) | 95.10 | 97.40 | 92.81 |
WSDAN (Inception) | 94.80 | 95.60 | 94.06 |
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN_d(VGG19) | 95.90 | 96.24 | 93.12 |
WSDAN(ResNet101) | 95.10 | 97.40 | 92.81 |
WSDAN_d(ResNet101) | 95.40 | 96.24 | 93.10 |
WSDAN (Inception) | 94.80 | 95.60 | 94.06 |
WSDAN_d (Inception) | 95.90 | 95.93 | 95.92 |
表4 增强图像后模型的训练结果与基础网络模型结果的对比 (%)
Tab.4 Comparison of training results of models with enhanced images and results of basic network models
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN_d(VGG19) | 95.90 | 96.24 | 93.12 |
WSDAN(ResNet101) | 95.10 | 97.40 | 92.81 |
WSDAN_d(ResNet101) | 95.40 | 96.24 | 93.10 |
WSDAN (Inception) | 94.80 | 95.60 | 94.06 |
WSDAN_d (Inception) | 95.90 | 95.93 | 95.92 |
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN(改进VGG19) | 95.00 | 95.57 | 95.62 |
表5 使用改进的VGG19网络的模型与使用基础VGG19网络的模型对比 (%)
Tab.5 Comparison of model with improved VGG19 network and model with basic VGG19 network
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN(改进VGG19) | 95.00 | 95.57 | 95.62 |
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(改进_1) | 92.60 | 93.12 | 92.18 |
WSDAN(改进_2) | 95.40 | 93.75 | 97.19 |
WSDAN(改进_3) | 96.40 | 97.81 | 94.99 |
WSDAN(改进_4) | 87.03 | 92.49 | 81.56 |
表6 增加不同的卷积层对比 (%)
Tab.6 Comparison of adding different convolutional layers
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(改进_1) | 92.60 | 93.12 | 92.18 |
WSDAN(改进_2) | 95.40 | 93.75 | 97.19 |
WSDAN(改进_3) | 96.40 | 97.81 | 94.99 |
WSDAN(改进_4) | 87.03 | 92.49 | 81.56 |
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN_d(改进VGG19) | 96.40 | 97.81 | 94.99 |
表7 增强图像结合改进网络的模型与基础网络模型的对比 (%)
Tab.7 Comparison of model with enhanced images combining improved network and basic network model
模型(特征网络) | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
WSDAN(VGG19) | 94.30 | 95.90 | 92.80 |
WSDAN_d(改进VGG19) | 96.40 | 97.81 | 94.99 |
模型 | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
VGG19 | 92.20 | 92.49 | 91.87 |
ResNet101 | 91.90 | 92.81 | 90.93 |
NTS_Net | 92.30 | 94.69 | 90.00 |
WSDAN | 94.30 | 95.90 | 92.80 |
本文方法 | 96.40 | 97.81 | 94.99 |
表8 不同分类网络的指标对比 (%)
Tab. 8 Comparison of indicators of different classification networks
模型 | 准确性 | 敏感性 | 特异性 |
---|---|---|---|
VGG19 | 92.20 | 92.49 | 91.87 |
ResNet101 | 91.90 | 92.81 | 90.93 |
NTS_Net | 92.30 | 94.69 | 90.00 |
WSDAN | 94.30 | 95.90 | 92.80 |
本文方法 | 96.40 | 97.81 | 94.99 |
1 | 叶玉如. 老年痴呆症——现代脑科学和医学研究面临的严峻挑战[J]. 生命科学, 2014, 26(1):1. (YE Y R. Alzheimer’s disease: a tough challenge faced by modern brain science and medical research[J]. Chinese Bulletin of Life Science, 2014, 26(1):1.) |
2 | BROOKMEYER R, JOHNSON E, ZIEGLER-GRAHAM K, et al. Forecasting the global burden of Alzheimer’s disease[J]. Alzheimer’s and Dementia, 2007, 3(3):186-191. 10.1016/j.jalz.2007.04.381 |
3 | 林伟铭,高钦泉,杜民. 卷积神经网络诊断阿尔兹海默症的方法[J].计算机应用, 2017, 37(12):3504-3508. 10.11772/j.issn.1001-9081.2017.12.3504 |
LIN W M, GAO Q Q, DU M. Convolutional neural network based method for the diagnosis of Alzheimer’s disease[J]. Journal of Computer Applications, 2017, 37(12):3504-3508. 10.11772/j.issn.1001-9081.2017.12.3504 | |
4 | YANG Z, LUO T G, WANG D, et al. Learning to navigate for fine-grained classification[C]// Proceedings of the 2018 European Conference on Computer Vision, LNCS11218. Chan: Springer, 2018:420-435. |
5 | LAM M, MAHASSENI B, TODOROVIC S. Fine-grained recognition as HSnet search for informative image parts[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017:6497-6506. 10.1109/cvpr.2017.688 |
6 | 边小勇,江沛龄,赵敏,等. 基于多分支神经网络模型的弱监督细粒度图像分类方法[J]. 计算机应用, 2020, 40(5):1295-1300. |
BIAN X Y, JIANG P L, ZHAO M, et al. Multi-branch neural network model based weakly supervised fine-grained image classification method[J]. Journal of Computer Applications, 2020, 40(5):1295-1300. | |
7 | 陆鑫伟,余鹏飞,李海燕,等. 基于注意力自身线性融合的弱监督细粒度图像分类算法[J]. 计算机应用, 2021, 41(5): 1319-1325. 10.1109/iaeac50856.2021.9390994 |
LU X W, YU P F, LI H Y, et al. Weakly supervised fine-grained image classification algorithm based on attention-attention bilinear pooling[J]. Journal of Computer Applications , 2021, 41(5): 1319-1325. 10.1109/iaeac50856.2021.9390994 | |
8 | XIAO T J, XU Y C, YANG K Y, et al. The application of two-level attention models in deep convolutional neural network for fine-grained image classification[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015:842-850. 10.1109/cvpr.2015.7298685 |
9 | JADERBERG M, SIMONYAN K, ZISSERMAN A, et al. Spatial transformer networks[C]// Proceedings of the 28th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2014: 2672-2680. 10.5244/c.28.88 |
10 | FU J L, ZHENG H L, MEI T, et al. Look closer to see better: recurrent attention convolutional neural network for fine-grained image recognition[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017:4476-4484. 10.1109/cvpr.2017.476 |
11 | HU T, QI H G, HUANG Q M, et al. See better before looking closer: weakly supervised data augmentation network for fine-grained visual classification[EB/OL]. (2019-03-23) [2021-04-19].. 10.1109/icme46284.2020.9102790 |
12 | 丁文谦,余鹏飞,李海燕,等. 基于Xception网络的弱监督细粒度图像分类[J/OL]. 计算机工程与应用. (2020-12-25) [2021-01-25]., |
YU P F, LI H Y, et al. Weakly supervised fine-grained image classification based on Xception network[J/OL]. Computer Engineering and Applications. (2020-12-25) [2021-01-25]. | |
13 | 李振东,钟勇,陈蔓,等. 角度余量损失和中心损失联合的深度人脸识别[J]. 计算机应用, 2019, 39(S2):55-58. 10.1109/icsidp47821.2019.9173230 |
LI Z D, ZHONG Y, CHEN M, et al. Deep face recognition combined with angular margin loss and center loss[J]. Journal of Computer Applications, 2019, 39(S2):55-58. 10.1109/icsidp47821.2019.9173230 | |
14 | 朱学玲,刘丽. 图像增强中的平滑滤波技术[J]. 科技信息, 2012(32):512. 10.3969/j.issn.1001-9960.2012.32.464 |
ZHU X L, LIU L. Smooth filtering technology in image enhancement[J]. Science and Technology Information, 2012(32):512. 10.3969/j.issn.1001-9960.2012.32.464 | |
15 | 郭红伟,余江,朱家兴,等. 基于局部直方图的加权均值滤波器[J]. 计算机应用, 2010, 30(11):3019-3021. 10.3724/sp.j.1087.2010.03019 |
GUO H W, YU J, ZHU J X, et al. Weighted mean filter based on local histogram[J]. Journal of Computer Applications, 2010, 30(11):3019-3021. 10.3724/sp.j.1087.2010.03019 | |
16 | SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL]. (2015-04-10) [2021-01-19].. 10.5244/c.28.6 |
17 | 颜晨欢,李白,章超杰,等. 一种基于VGG的手写字母识别算法[J]. 信息技术与信息化, 2020(12):63-65. 10.3969/j.issn.1672-9528.2020.12.019 |
YAN C H, LI B, ZHANG C J, et al. A handwritten letter recognition algorithm based on VGG[J]. Information Technology and Informatization, 2020(12):63-65. 10.3969/j.issn.1672-9528.2020.12.019 | |
18 | 王羽徵,程远,毕海,等. 基于深度学习VGG网络的海洋单细胞藻类识别算法研究[J]. 大连海洋大学学报, 2021, 36(2):334-339. |
WANG Y Z, CHENG Y, BI H, et al. Recognition algorithm of marine single-cell algae based on deep learning VGG network[J]. Journal of Dalian Ocean University, 2021, 36(2):334-339. | |
19 | 陈津徽,张元良,尹泽睿. 基于改进的VGG19网络的面部表情识别[J]. 电脑知识与技术, 2020, 16(29):187-188. 10.1117/12.2574468 |
CHEN J H, ZHANG Y L, YIN Z R. Facial expression recognition based on improved VGG19 network[J]. Computer Knowledge and Technology, 2020, 16(29):187-188. 10.1117/12.2574468 | |
20 | SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2017:618-626. 10.1109/iccv.2017.74 |
21 | CHATTOPADHAY A, SARKAR A, HOWLADER P, et al. Grad-CAM++: generalized gradient-based visual explanations for deep convolutional networks[C]// Proceedings of the 2018 IEEE Winter Conference on Applications of Computer Vision. Piscataway: IEEE, 2018:839-847. 10.1109/wacv.2018.00097 |
22 | 陆小玲,吴海锋,曾玉,等. 3D迁移网络的阿尔茨海默症分类研究[J]. 计算机工程与应用, 2021, 57(16):253-262. 10.1109/icccr49711.2021.9349393 |
LU X L, WU H F, ZENG Y, et al. 3D transfer learning network for classification of Alzheimer’s disease[J]. Computer Engineering and Applications, 2021, 57(16):253-262. 10.1109/icccr49711.2021.9349393 | |
23 | 张柏雯,林岚,吴水才. 基于AlexNet模型的AD分类[J]. 北京工业大学学报, 2020, 46(1):68-74. 10.11936/bjutxb2018070029 |
ZHANG B W, LIN L, WU S C. Efficient Alzheimer’s disease classification based on AlexNet model[J]. Journal of Beijing University of Technology, 2020, 46(1):68-74. 10.11936/bjutxb2018070029 |
[1] | 杨莹, 郝晓燕, 于丹, 马垚, 陈永乐. 面向图神经网络模型提取攻击的图数据生成方法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2483-2492. |
[2] | 汪炅, 唐韬韬, 贾彩燕. 无负采样的正样本增强图对比学习推荐方法PAGCL[J]. 《计算机应用》唯一官方网站, 2024, 44(5): 1485-1492. |
[3] | 朱子蒙, 李志新, 郇战, 陈瑛, 梁久祯. 基于三元中心引导的弱监督视频异常检测[J]. 《计算机应用》唯一官方网站, 2024, 44(5): 1452-1457. |
[4] | 郭洁, 林佳瑜, 梁祖红, 罗孝波, 孙海涛. 基于知识感知和跨层次对比学习的推荐方法[J]. 《计算机应用》唯一官方网站, 2024, 44(4): 1121-1127. |
[5] | 郭安迪, 贾真, 李天瑞. 基于伪实体数据增强的高精准率医学领域实体关系抽取[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 393-402. |
[6] | 党伟超, 张磊, 高改梅, 刘春霞. 融合片段对比学习的弱监督动作定位方法[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 548-555. |
[7] | 宋逸飞, 柳毅. 基于数据增强和标签噪声的快速对抗训练方法[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3798-3807. |
[8] | 胡新荣, 陈静雪, 黄子键, 王帮超, 姚迅, 刘军平, 朱强, 杨捷. 基于图卷积网络的掩码数据增强[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3335-3344. |
[9] | 王强, 黄小明, 佟强, 刘秀磊. 基于边界框标注的弱监督显著性目标检测算法[J]. 《计算机应用》唯一官方网站, 2023, 43(6): 1910-1918. |
[10] | 蔡引江, 许光俊, 马喜波. 图结构表示下的药物数据增强方法[J]. 《计算机应用》唯一官方网站, 2023, 43(4): 1136-1141. |
[11] | 林呈宇, 王雷, 薛聪. 标签语义增强的弱监督文本分类模型[J]. 《计算机应用》唯一官方网站, 2023, 43(2): 335-342. |
[12] | 孙邱杰, 梁景贵, 李思. 基于BART噪声器的中文语法纠错模型[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 860-866. |
[13] | 胡聪, 华钢. 基于注意力机制的弱监督动作定位方法[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 960-967. |
[14] | 曹一珉, 蔡磊, 高敬阳. 基于生成对抗网络的基因数据生成方法[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 783-790. |
[15] | 彭禹, 宋耀莲, 杨俊. 基于数据增强的运动想象脑电分类[J]. 《计算机应用》唯一官方网站, 2022, 42(11): 3625-3632. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||