基于残差混合注意力机制的脑部CT图像分类卷积神经网络模型

doi:10.12263/DZXB.20200881

电子学报 ›› 2021, Vol. 49 ›› Issue (5): 984-991.DOI: 10.12263/DZXB.20200881

基于残差混合注意力机制的脑部CT图像分类卷积神经网络模型

乔思波¹, 庞善臣¹, 王敏², 翟雪¹, 于世行³, 丁桐⁴

1. 中国石油大学(华东)计算机科学与技术学院, 山东青岛 266580;
2. 中国石油大学(华东)控制科学与工程学院, 山东青岛 266580;
3. 天津工业大学机械工程学院, 天津 300387;
4. 山东大学软件学院, 山东济南 250101

收稿日期:2020-08-13 修回日期:2020-09-25 出版日期:2021-05-25 发布日期:2021-05-25
通讯作者: 庞善臣(通信作者) 男,1974年10月生,山东济宁人.现为中国石油大学(华东)计算机科学与技术学院三级教授、博士生导师,山东省泰山产业领军人才.主要从事Petri网理论、可信计算、人工智能算法、医疗图像处理等方面的研究. E-mail:pangsc@upc.edu.cn
作者简介:乔思波 男,1993年7月生,山东聊城人.现为中国石油大学(华东)计算机技术与资源信息工程专业博士研究生,主要研究方向为医疗图像处理、人工智能算法及其应用. E-mail:siboqiao@qq.com;王敏女,1994年12月生,江苏泰州人.现为中国石油大学(华东)控制科学与工程专业博士研究生.主要研究方向为图像处理;翟雪女,1995年8月生,山东聊城人.现为中国石油大学(华东)计算机技术与资源信息工程专业博士研究生.主要研究方向为区块链、图像分析;于世行 男,1992年11月生,山东聊城人.现为天津工业大学机械工程专业博士研究生.主要研究方向为数据分析、图像处理和推荐系统;丁桐男,1994年7月生,宁夏银川人.现为山东大学人工智能专业博士研究生.主要研究方向为人工智能算法及其应用、医疗图像处理.
基金资助:
山东省重大科技创新工程（No.2019TSLH0214）；泰山产业领军人才（No.tscy20180416）；国家自然科学基金（No.61873281，No.61972416）

A Convolutional Neural Network for Brain CT Image Classification Based on Residual Hybrid Attention Mechanism

QIAO Si-bo¹, PANG Shan-chen¹, WANG Min², ZHAI Xue¹, YU Shi-hang³, DING Tong⁴

1. College of Computer Science and Technology, China University of Petroleum, Qingdao, Shandong 266580, China;
2. College of Control Science and Engineering, China University of Petroleum, Qingdao, Shandong 266580, China;
3. College of Mechanical Engineering, Tiangong University, Tianjin 300387, China;
4. College of Software, Shandong University, Jinan, Shandong 250101, China

Received:2020-08-13 Revised:2020-09-25 Online:2021-05-25 Published:2021-05-25

摘要/Abstract

摘要： 针对阿尔兹海默症、病变(如脑肿瘤)和健康老化的3类脑部CT图像分类问题,本文提出了一种改进的ResNet-10卷积神经网络模型.该模型在网络的残差映射结构中加入残差混合注意力模块,解决了原模型提取的特征分辨性弱的问题,精确捕捉了脑部组织在CT图像中的位置和内容信息;此外,本文设计了全局平均池化层,简化了模型的复杂度,并在其后引入Dropout机制,缓解了过拟合.在训练阶段,该模型建立了标签平滑交叉熵损失函数,使模型在样本数量有限的情况下仍有较强的泛化能力.系列实验证明了改进后的ResNet-10网络模型在分类脑部CT图像时达到97.47%的分类精度.

关键词: 残差混合注意力模块, 标签平滑, 脑部CT, 卷积神经网络

Abstract: To classify three types of brain CT(computerized tomography) images in Alzheimer's disease,lesion (e.g.,brain tumor) and healthy aging,an improved ResNet-10 convolutional neural network is proposed in this papers.A residual hybrid attention module is embedded in the residual identity mapping to capture the location and content information of brain tissue in brain CT images,solving the original model to extract weak distinguish features problems.In addition,to simplify the improved model and alleviate the overfitting,several techniques such as global average pooling and Dropout are used in the model.Moreover,to have strong generalization ability in the case of limited sample quantity,tag smoothing cross-entropy loss function is adopted to train the model.Experimental results show that the improved ResNet-10 achieves 97.47% accuracy in classifying brain CT images.

Key words: residual hybrid attention module, label smoothing, brain CT, convolutional neural network

中图分类号:

TP399

乔思波, 庞善臣, 王敏, 翟雪, 于世行, 丁桐. 基于残差混合注意力机制的脑部CT图像分类卷积神经网络模型[J]. 电子学报, 2021, 49(5): 984-991.

QIAO Si-bo, PANG Shan-chen, WANG Min, ZHAI Xue, YU Shi-hang, DING Tong. A Convolutional Neural Network for Brain CT Image Classification Based on Residual Hybrid Attention Mechanism[J]. Acta Electronica Sinica, 2021, 49(5): 984-991.

参考文献

[1] 王旭,段辉宏,聂生东.基于CT影像组学的非小细胞肺癌预后分析方法[J].电子学报,2020,48(4):637-642. Wang X,Duan H H,Nie S D.Prognostic analysis method for non-small cell lung cancer based on CT radiomics[J].Acta Electronica Sinica,2020,48(4):637-642.(in Chinese).
[2] 桂春燕.应用CT和磁共振成像诊断盆腔肿瘤的临床价值[J].影像研究与医学应用,2020,4(15):40-41. Gui C Y.The clinical value of CT and magnetic resonance imaging in the diagnosis of pelvic tumors[J].Journal of Imaging Research and Medical Applications,2020,4(15):40-41.(in Chinese).
[3] 朱立军.CT、MRI进行股骨头坏死放射诊断的效果观察[J].影像研究与医学应用,2020,4(15):136-137. Zhu L J.The effect of CT and MRI in radiographic diagnosis of femoral head necrosis[J].Journal of Imaging Research and Medical Applications,2020,4(15):136-137.(in Chinese).
[4] Gao W X,Hui R,Tian Z. Classification of CT brain images based on deep learning networks[J].Computer Methods and Programs in Biomedicine,2017,138:49-56.
[5] Cheng J Z,Ni D,Chou Y H,et al.Computer-aided diagnosis with deep learning architecture:Applications to breast lesions in US images and pulmonary nodules in CT scans[J].Scientific Reports.2016,6:1-13.
[6] 王磊,徐涛,宋传东,等.基于深度学习的miRNA与疾病相关性预测算法[J].电子学报,2020,48(5):870-877. Wang L,Xu T,Song C D,et al.Prediction algorithm of association between miRNAs and diseases based on deep learning[J].Acta Electronica Sinica,2020,48(5):870-877.(in Chinese).
[7] Ciompi F,Hoop D B,Riel S J,et al.Automatic classification of pulmonary peri-fissural nodules in computed tomography using an ensemble of 2D views and a convolutional neural network out-of-the-box[J].Medical Image Analysis,2015,26(1):195-202.
[8] Oktay O,Ferrante E,Kamnitsas K,et al.Anatomically constrained neural networks (ACNNs):Application to cardiac image enhancement and segmentation[J].IEEE Transactions on Medical Imaging,2018,37(2):384-395.
[9] Leclerc S,Smistad E,Pedrosa J,et al.Deep learning for segmentation using an open large-scale dataset in 2D echocardiography[J].IEEE Transactions on Medical Imaging,2019,38(9):2198-2210.
[10] Pereira S,Pinto A,Alves V,et al.Brain tumor segmentation using convolutional neural networks in MRI images[J].IEEE Transactions on Medical Imaging,2016,35(5):1240-1251.
[11] Kumar A,Kim J,Lyndon D,et al.An ensemble of fine-tuned convolutional neural networks for medical image classification[J].IEEE Journal of Biomedical and Health Informatics,2017,21(1):31-40.
[12] Farooq A,Anwar S,Awais M,et al.A deep CNN based multi-class classification of Alzheimer's disease using MRI[A].IEEE International Conference on Imaging Systems and Techniques[C].New York,USA:IEEE,2017.1-6.
[13] Zahedinasab R,Mohseni H.Using deep convolutional neural networks with adaptive activation functions for medical CT brain image classification[A].International Iranian Conference on Biomedical Engineering[C].New York,USA:IEEE,2018.1-6.
[14] Wang F,Jiang M Q,Qian C,et al.Residual attention network for image classification[A].IEEE Conference on Computer Vision and Pattern Recognition[C].New York,USA:IEEE,2017.3156-3164.
[15] Hu J,Shen L,Sun G.Squeeze-and-Excitation networks[A].IEEE Conference on Computer Vision and Pattern Recognition[C].New York,USA:IEEE,2018.7132-7141.
[16] Woo S,Park J,Lee J,et al.CBAM:Convolutional block attention module[A].Proceedings of the European Conference on Computer Vision[C].Berlin,Germany:Springer,2018.3-19.
[17] Cao Y H,Chen K,Loy C C,et al.Prime sample attention in object detection[A].IEEE Conference on Computer Vision and Pattern Recognition[C].New York,USA:IEEE,2020.11583-11591.
[18] Kaul C,Manandhar S,Pears N.Focusnet:An attention-based fully convolutional network for medical image segmentation[A].IEEE 16th International Symposium on Biomedical Imaging[C].New York,USA:IEEE,2019.455-458.
[19] 唐海桃,薛嘉宾,韩纪庆.一种多尺度前向注意力模型的语音识别方法[J].电子学报.2020,48(7):1276-1283. Tang H T,Xue J B,Han J Q,et al.A method of multi-scale forward attention model for speech recognition[J].Acta Electronica Sinica,2020,48(7):1276-1283.(in Chinese).
[20] He K,Zhang X,Ren S,et al.Deep residual learning for image recognition[A].IEEE Conference on Computer Vision and Pattern Recognition[C].New York,USA:IEEE,2016.770-778.
[21] Xiang L,Shuo C,Xiao L H,et al.Understanding the disharmony between dropout and batch normalization by variance shift[A].IEEE Conference on Computer Vision[C].New York,USA:IEEE,2019.2677-2685.
[22] Szegedy C,Vanhoucke V,Ioffe S,et al.Rethinking the inception architecture for computer vision[A].IEEE Conference on Computer Vision and Pattern Recognition[C].New York,USA:IEEE,2016.2818-2826.
[23] Kingma D,Ba J.Adam:A method for stochastic optimization[J].arXiv Preprint,2015,arXiv:1412.6980.

基于残差混合注意力机制的脑部CT图像分类卷积神经网络模型

A Convolutional Neural Network for Brain CT Image Classification Based on Residual Hybrid Attention Mechanism

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	廖勇, 李玉杰. 一种轻量化低复杂度的FDD大规模MIMO系统CSI反馈方法[J]. 电子学报, 2022, 50(5): 1211-1217.
[2]	张云, 化青龙, 姜义成, 徐丹. 基于混合型复数域卷积神经网络的三维转动舰船目标识别[J]. 电子学报, 2022, 50(5): 1042-1049.
[3]	崔亚奇, 何友, 唐田田, 熊伟. 一种深度学习航迹关联方法[J]. 电子学报, 2022, 50(3): 759-763.
[4]	伍邦谷, 张苏林, 石红, 朱鹏飞, 王旗龙, 胡清华. 基于多分支结构的不确定性局部通道注意力机制[J]. 电子学报, 2022, 50(2): 374-382.
[5]	李居朋, 王颖慧, 李刚. 医学图像关键点检测深度学习方法研究与挑战[J]. 电子学报, 2022, 50(1): 226-237.
[6]	赵琰, 赵凌君, 匡纲要. 基于注意力机制特征融合网络的SAR图像飞机目标快速检测[J]. 电子学报, 2021, 49(9): 1665-1674.
[7]	管永明, 王刚, 骆凯波, 吕梁, 吕晓雯, 史玉良. 基于极端不平衡学习的泛化低压异常箱表关系识别研究与应用[J]. 电子学报, 2021, 49(8): 1507-1514.
[8]	徐频捷, 陈逸杰, 李之南, 赵地. 基于事件驱动的车道线识别算法研究[J]. 电子学报, 2021, 49(7): 1379-1385.
[9]	罗会兰, 袁璞, 童康. 基于深度学习的显著性目标检测方法综述[J]. 电子学报, 2021, 49(7): 1417-1427.
[10]	程旭, 宋晨, 史金钢, 周琳, 张毅锋, 郑钰辉. 基于深度学习的通用目标检测研究综述[J]. 电子学报, 2021, 49(7): 1428-1438.
[11]	张昱, 刘开峰, 张全新, 王艳歌, 高凯龙. 基于组合-卷积神经网络的中文新闻文本分类[J]. 电子学报, 2021, 49(6): 1059-1067.
[12]	裴翰奇, 杨春玲, 魏志超, 曹燕. 基于SPL迭代思想的图像压缩感知重构神经网络[J]. 电子学报, 2021, 49(6): 1195-1203.
[13]	刘杰, 葛一凡, 田明, 马力强. 基于ZYNQ的可重构卷积神经网络加速器[J]. 电子学报, 2021, 49(4): 729-735.
[14]	吴俊劼, 陈震, 张聪炫, 江少锋, 尚璇. 基于特征级联卷积网络的双目立体匹配[J]. 电子学报, 2021, 49(4): 690-695.
[15]	李志欣, 凌锋, 张灿龙, 马慧芳. 融合两级相似度的跨媒体图像文本检索[J]. 电子学报, 2021, 49(2): 268-274.