面向白内障识别的临床特征校准注意力网络

doi:10.3778/j.issn.1002-8331.2210-0158

摘要/Abstract

摘要： 近年来，卷积神经网络（convolutional neural network，CNN）模型已经被广泛用于年龄相关性白内障自动分类任务，然而，鲜有研究工作将年龄相关性白内障的临床先验知识注入卷积神经网络架构设计中，以此来提高年龄相关性白内障的分类效果和改善模型决策过程的可解释性。提出了一种临床特征校准注意力网络（clinical feature recalibration attention network，CFANet）模型用于自动识别年龄相关性白内障严重级别。在CFANet中，设计了一个简单且有效的临床特征校准注意力模块（clinical feature recalibration attention block，CFA），其不仅能对不同临床特征类型进行自适应地加权融合，还通过门控操作符来突出重要通道和抑制不重要通道。在一个核性白内障的眼前节光学相干断层成像影像（anterior segment optical coherence tomography，AS-OCT）数据集和一个公开眼科影像数据集上进行了充分实验，实验结果表明，相较于squeeze-and-excitation network（SENet）、efficient channel network(ECANet)、style-based recalibration module（SRM），CFANet在AS-OCT数据集上的分类准确率至少提升了3.54个百分点，同时在公开的眼科影像数据集上的分类结果比先进的神经网络模型和已发表的研究工作提升了1个百分点以上。此外，还通过可视化方法分析临床特征的权重分布和通道的注意力权重分布来提高该文模型决策过程的可解释性。

Abstract: In recent years, convolutional neural networks (CNNs) have been widely used for automatic age-related cataract classification. However, incorporating clinical prior knowledge of age-related cataracts into CNN design improves the classification performance and the interpretability of the decision-making process of CNNs, which has been less studied. To this problem, this paper proposes a clinical-feature recalibration attention network (CFANet) to classify age-related cataract severity levels automatically. In the CFANet, a simple yet effective clinical feature recalibration attention (CFA) block is designed to fuse clinical features adaptively by setting relative weights, aiming to highlight significant channels and suppress redundant ones. This paper conducts extensive experiments on a clinical AS-OCT image dataset of nuclear cataract and a public eye image dataset to verify the effectiveness of CFANet. The results show that CFANet outperforms advanced baselines by above 3.54 percentage points of accuracy on the clinical AS-OCT image dataset, such as squeeze-and-excitation network (SENet), efficient channel network (ECANet), style-based recalibration module (SRM). And the results on the public eye dataset also show that compared with strong attention-based CNNs and published works, proposed method obtains over 1 percentage point improvement. Moreover, this paper also uses visualization methods to analyze clinical feature weights and channel attention weights to enhance the interpretability of the decision-making process for proposed method.

Key words: age-related cataract classification, anterior segment optical coherence tomography (AS-OCT), clinical-feature recalibration attention, interpretability, convolutional neural network (CNN), visualization

章晓庆, 肖尊杰, 赵宇航, 巫晓, 东田理沙, 刘江. 面向白内障识别的临床特征校准注意力网络[J]. 计算机工程与应用, 2024, 60(3): 321-330.

ZHANG Xiaoqing, XIAO Zunjie, ZHAO Yuhang, WU Xiao, Risa Higashita, LIU Jiang. Clinical Feature Recalibration Attention Network for Cataract Recognition[J]. Computer Engineering and Applications, 2024, 60(3): 321-330.

参考文献

[1] WANG W, YAN W, FOTIS K, et al. Cataract surgical rate and socioeconomics: a global study[J]. Investigative Opthalmology & Visual Science, 2016, 57(14): 5872-5881.
[2] GALI H E, SELLA R, AFSHARI N A. Cataract grading systems: a review of past and present[J]. Current Opinion in Ophthalmology, 2019, 30(1): 13-18.
[3] OZGOKCE M, BATUR M, ALPASLAN M, et al. A comparative evaluation of cataract classifications based on shear-wave elastography and B-mode ultrasound findings[J]. Journal of Ultrasound, 2019, 22(4): 447-452.
[4] CHYLACKL T, WOLFEJ K, SINGERD M, et al. The lens opacities classification system III[J]. Archives of Ophthalmology, 1993, 111(6): 831-836.
[5] GALI H E, SELLA R, AFSHARI N A. Cataract grading systems: a review of past and present[J]. Current Opinion in Ophthalmology, 2019, 30(1): 13-18.
[6]李建强, 张苓琳, 张莉, 等.基于深度学习的白内障识别与分级[J].第二军医大学学报, 2018, 39(8): 878-885.
LI J Q, ZHANG L L, ZHANG L, et al. Cataract recognition and grading based on deep learning[J]. Academic Journal of Second Military Medical University, 2018, 39(8): 878-885.
[7] XU X, ZHANG L, LI J, et al. A hybrid global-local representation CNN model for automatic cataract grading[J]. IEEE Journal of Biomedical and Health Informatics, 2020, 24(2): 556-567.
[8] 刘振宇, 宋建聪.基于深度学习的白内障自动诊断方法研究[J].微处理机, 2019, 40(3): 48-52.
LIU Z Y, SONG J C. Research on automatic diagnosis method of cataract based on deep learning[J]. Microprocessors, 2019, 40(3): 48-52.
[9] WONG A L, LEUNG K S, WEINREB R N, et al. Quantitative assessment of lens opacities with anterior segment optical coherence tomography[J]. British Journal of Ophthalmology, 2009, 93(1): 61-65.
[10] HOU Q, ZHOU D, FENG J. Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 13713-13722.
[11] HUANG Z, WANG X, HUANG L, et al. Ccnet: criss-cross attention for semantic segmentation[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019: 603-612.
[12] HU J, SHEN L, SUN G, et al. Squeeze-and-excitation networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018.
[13] GU, J X, WANG Z H, KUEN J, et al. Recent advances in convolutioal neural networks[J]. Pattern Recognition, 2018, 77: 354-377.
[14] DOS SANTOS V A, SCHMETTERER L, STEGMANN H, et al. CorneaNet: fast segmentation of cornea OCT scans of healthy and keratoconic eyes using deep learning[J]. Biomedical Optics Express, 2019, 10(2): 622.
[15] BRENTON K, MARK D, GAO T, et al. Real-time corneal segmentation and 3D needle tracking in intrasurgical OCT[J]. Biomedical Optics Express, 2018, 9(6): 2716-2732.
[16] FU H Z, BASKARAN M, XU Y W, et al. A deep learning system for automated angle-closure detection in anterior segment optical coherence tomography images[J]. American Journal of Ophthalmology, 2019, 203: 37-45.
[17] FU H, XU Y, LIN S, et al. Multi-context deep network for angle-closure glaucoma screening in anterior segment OCT[J].arXiv:1809.03239, 2018.
[18] CHEN D, LI Z, HUANG J, et al. Lens nuclear opacity quantitation with long-range swept-source optical coherence tomography: correlation to LOCS III and a scheimpflug imaging-based grading system[J]. British Journal of Ophthalmology, 2018, 103(8).
[19] ALBERTO D C, ANTONIO B, SILVESTRE M, et al. Three-dimensional cataract crystalline lens imaging with swept-source optical coherence tomography[J]. Investigative Opthalmology & Visual Science, 2018, 59(2): 897-903.
[20] MAKHOTKINA N Y, BERENDSCHOT T T J M, VAN DEN BIGGELAAR F J H M, et al. Comparability of subjective and objective measurements of nuclear density in cataract patients[J]. Acta Ophthalmologica, 2018, 96(4): 356-363.
[21] WANG W, ZHANG J, GU X, et al. Objective quantification of lens nuclear opacities using swept-source anterior segment optical coherence tomography[J]. British Journal of Ophthalmology, 2022, 106(6): 790-794.
[22] CAO G, ZHAO W, HIGASHITA R, et al. An efficient lens structures segmentation method on AS-OCT images[C]// 2020 42nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) in conjunction with the 43rd Annual Conference of the Canadian Medical and Biological Engineering Society, 2020.
[23] ZHANG X, XIAO Z, HIGASHITA R, et al. A novel deep learning method for nuclear cataract classification based on anterior segment optical coherence tomography images[C]//2020 IEEE International Conference on Systems, Man, and Cybernetics (IEEE SMC), 2020.
[24] 章晓庆, 方建生, 肖尊杰, 等.基于眼前节相干光断层扫描成像的核性白内障分类算法[J].计算机科学, 2022, 49(3): 204-210.
ZHANG X Q, FANG J S, XIAO J J, et al. Classification algorithm of nuclear cataract based on anterior segment coherence tomography image[J].Computer Science, 2022, 49(3): 204-210.
[25] XIAO Z, ZHANG X, HIGASHITA R, et al. Gated channel attention network for cataract classification on AS-OCT image[C]//International Conference on Neural Information Processing. Cham: Springer, 2021: 357-368.
[26] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//European Conference on Computer Vision.Cham: Springer, 2018.
[27] WANG X, GIRSHICK R, GUPTA A, et al. Non-local neural networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 7794-7803.
[28] 陈朝一, 许波, 吴英, 等. 医学图像处理中的注意力机制研究综述[J]. 计算机工程与应用, 2022, 58(5): 23-33.
CHEN C Y, XU B, WU Y, et al. Overview of research on attention mechanism in medical image processing[J]. Computer Engineering and Applications, 2022, 58(5): 23-33.
[29] 王静, 孙紫雲, 郭苹, 等. 改进YOLOv5的白细胞检测算法[J]. 计算机工程与应用, 2022, 58(4): 134-142.
WANG J, SUN Z Y, GUO P, et al. Improved leukocyte detection algorithm of YOLOv5[J]. Computer Engineering and Applications, 2022, 58(4): 134-142.
[30] WANG Q, WU B, ZHU P, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]//CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020.
[31] LEE H J, KIM H E, NAM H. SRM: a style-based recalibration module for convolutional neural networks[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019: 1854-1862.
[32] PARK J, WOO S, LEE J Y, et al. A simple and light-weight attention module for convolutional neural networks[J]. International Journal of Computer Vision, 2020, 128(4): 783-798.
[33] IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]//International Conference on Machine Learning, 2015: 448-456.
[34] 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, 2016: 770-778.
[35] KERMANY D S, GOLDBAUM M, CA I W, et al. Identifying medical diagnoses and treatable diseases by image-based deep learning[J]. Cell, 2018, 172(5): 1122-1131.
[36] DAS V, DANDAPAT S, BORA P K. Multi-scale deep feature fusion for automated classification of macular pathologies from OCT images[J]. Biomedical Signal Processing and Control, 2019, 54: 101605.
[37] CAO L, LI H, ZHANG Y, et al. Hierarchical method for cataract grading based on retinal images using improved Haar wavelet[J]. Information Fusion, 2019, 53: 196-208.
[38] ZHANG X, HU Y, FANG J S, et al. Machine learning for cataract classification and grading on ophthalmic imaging modalities: a survey[J]. arXiv:2012.04830, 2020.
[39] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv:1409. 1556, 2014.
[40] FANG L, WANG C, LI S, et al. Attention to lesion: lesion-aware convolutional neural network for retinal optical coherence tomography image classification[J]. IEEE Transactions on Medical Imaging, 2019, 38(8): 1959-1970.
[41] ZHOU B, KHOSLA A, LAPEDRIZA A, et al. Learning deep features for discriminative localization[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
[42] THYLEFORS B, CHYLACK L T, KONYAMA K, et al. A simplified cataract grading system the WHO cataract grading group[J]. Ophthalmic Epidemiol, 2002, 9(2): 83-95.