LUO Yan, WANG Cong, YE Wenling. An Interpretable Prediction Model for Acute Kidney Injury Based on XGBoost and SHAP[J]. Journal of Electronics & Information Technology, 2022, 44(1): 27-38. doi: 10.11999/JEIT210931
Citation:
ZHUO Li, SUN Liangliang, ZHANG Hui, LI Xiaoguang, ZHANG Jing. TCM Tongue Color Classification Method under Noisy Labeling[J]. Journal of Electronics & Information Technology, 2022, 44(1): 89-98. doi: 10.11999/JEIT210935
LUO Yan, WANG Cong, YE Wenling. An Interpretable Prediction Model for Acute Kidney Injury Based on XGBoost and SHAP[J]. Journal of Electronics & Information Technology, 2022, 44(1): 27-38. doi: 10.11999/JEIT210931
Citation:
ZHUO Li, SUN Liangliang, ZHANG Hui, LI Xiaoguang, ZHANG Jing. TCM Tongue Color Classification Method under Noisy Labeling[J]. Journal of Electronics & Information Technology, 2022, 44(1): 89-98. doi: 10.11999/JEIT210935
Tongue color is one of the most concerned diagnostic features of tongue diagnosis in Traditional Chinese Medicine (TCM). Automatic and accurate tongue color classification is an important content of the objectification of tongue diagnosis. Due to the vagueness of the visual boundaries between different types of tongue colors and the subjectivity of the doctors, the annotated tongue image data samples often contain noises, which has a negative effect on the training of the tongue color classification model. Therefore, in this paper, a tongue color classification method in TCM with noisy labels is proposed. Firstly, a two-stage data cleaning method is proposed to identify and clean noisy labeled samples. Secondly, a lightweight Convolutional Neural Network (CNN) based on the channel attention mechanism is designed in this paper to achieve accurate classification of tongue color by enhancing the expressiveness of features. Finally, a knowledge distillation strategy with a noise sample filtering mechanism is proposed. This strategy adds a noise sample filtering mechanism led by the teacher network to eliminate further noise samples. At the same time, the teacher network is used to guide the training of the light convolutional neural network to improve the classification performance.The experimental results on the self-established TCM tongue color classification dataset show that the proposed method in this paper can significantly improve the classification accuracy with lower computational complexity, reaching 93.88%.
疾病的早期预测可以在支持医疗卫生专业人员方面发挥重要作用,据统计,11%的医院死亡是由于未能及时识别和治疗病情恶化所致[28]。近5年来,ML方法在准确、及时预测AKI高危患者方面发挥了重要作用。例如,Flechet等人[29]开发并验证了随机森林预测模型,成功预测了成年ICU患者的AKI,受试者工作特征曲线下面积(Area Under the receiver operating characteristics Curve, AUC)达到0.84。
如前所述,大多数模型的AUC表现一般[30,32-35],而一些研究集中于特定的患者群体,如心脏手术患者[31],或研究样本量小[37],从而限制了这些模型的使用。此外,从表1可以看出,目前大多数研究中使用的特征向量非常繁琐,收集和计算也很复杂。这些研究大多缺乏模型可解释性,只有有限的解释提供简单的特征重要性结果。相比之下,本文在前期研究的基础上,使用公开可访问的包含超过46000名患者的去识别健康数据的重症监护医学信息数据库MIMIC(Medical Information Mart for Intensive Care)III进行数据分析和模型开发[38],最后基于XGBoost算法构建了重症监护病房患者的AKI早期预测模型,并比较了XGBoost与其他4种流行的机器学习技术的性能。模型中仅使用常见的生命体征和实验室检测指标,通过有效的数据预处理和XGBoost模型参数调整,取得了良好的AKI早期风险预测性能。然后,利用SHAP 估计的Shapley值从全局和局部两个角度对预测模型进行解释。解释结果不依赖所使用的预测模型,这保证了结果的可靠性并为解决临床问题提供更多的证据支撑。这些成为这项工作的主要贡献。
LI Naimin. Complete Collection of Tongue Diagnosis of Traditional Chinese Medicine[M]. Beijing: Academy Press, 1994: 30–170.
[2]
WANG Xingyuan and WANG Zongyu. A novel method for image retrieval based on structure elements’ descriptor[J]. Journal of Visual Communication and Image Representation, 2013, 24(1): 63–74. doi: 10.1016/j.jvcir.2012.10.003
[3]
WANG Xingyuan, CHEN Zhifeng, and YUN Jiaojiao. An effective method for color image retrieval based on texture[J]. Computer Standards & Interfaces, 2012, 34(1): 31–35. doi: 10.1016/j.csi.2011.05.001
[4]
WANG Chunpeng, WANG Xingyuan, XIA Zhiqiu, et al. Image description with polar harmonic Fourier moments[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2020, 30(12): 4440–4452. doi: 10.1109/TCSVT.2019.2960507
[5]
XIA Zhiqiu, WANG Xingyuan, ZHOU Wenjie, et al. Color medical image lossless watermarking using chaotic system and accurate quaternion polar harmonic transforms[J]. Signal Processing, 2019, 157: 108–118. doi: 10.1016/j.sigpro.2018.11.011
[6]
WANG Xingyuan and WANG Zongyu. The method for image retrieval based on multi-factors correlation utilizing block truncation coding[J]. Pattern Recognition, 2014, 47(10): 3293–3303. doi: 10.1016/j.patcog.2014.04.020
WANG Yiran and ZHANG Xinfeng. Tongue color classification based on AdaBoost cascade framework[J]. Beijing Biomedical Engineering, 2020, 39(1): 8–14. doi: 10.3969/j.issn.1002-3208.2020.01.002
ZHANG Yifan, HU Guangqin, and ZHANG Xinfeng. An algorithm study on tongue color recognition of patients with acne based on support vector machine[J]. Beijing Biomedical Engineering, 2016, 35(1): 7–11. doi: 10.3969/j.issn.1002-3208.2016.01.02
ZHANG Jing, ZHANG Xinfeng, WANG Yazhen, et al. Research on multi-label learning in the classification of tongue images in TCM[J]. Beijing Biomedical Engineering, 2016, 35(2): 111–116. doi: 10.3969/j.issn.1002-3208.2016.02.01
WANG Yonggang, YANG Jie, and ZHOU Yue. Tongue image color recognition in traditional Chinese medicine[J]. Journal of Biomedical Engineering, 2005, 22(6): 1116–1120. doi: 10.3321/j.issn:1001-5515.2005.06.008
WANG Aimin, ZHAO Zhongxu, and SHEN Lansun. Research on the tongue color classification in automatic tongue analysis of traditional Chinese medicine[J]. Beijing Biomedical Engineering, 2000, 19(3): 136–142. doi: 10.3969/j.issn.1002-3208.2000.03.002
[12]
Hou Jun, Su Hongyi, Yan Bo, etal. Classification of tongue color based on CNN[C]. Proceedings of the 2017 IEEE 2nd International Conference on Big Data Analysis, Beijing, China, 2017: 725–729.
[13]
FU Shengyu, ZHENG Hong, YANG Zijiang, etal. Computerized tongue coating nature diagnosis using convolutional neural network[C]. The 2017 IEEE 2nd International Conference on Big Data Analysis, Beijing, China, 2017: 730–734.
XIAO Qingxin. Research on TCM tongue feature classification technology based on deep learning[D]. [Master dissertation], Beijing University of Technology, 2019.
[15]
UNAR S, WANG Xingyuan, and ZHANG Chuan. Visual and textual information fusion using Kernel method for content based image retrieval[J]. Information Fusion, 2018, 44: 176–187. doi: 10.1016/j.inffus.2018.03.006
[16]
UNAR S, WANG Xingyuan, WANG Chunpeng, et al. A decisive content based image retrieval approach for feature fusion in visual and textual images[J]. Knowledge-Based Systems, 2019, 179: 8–20. doi: 10.1016/j.knosys.2019.05.001
[17]
PENG Xiaojiang, WANG Kai, ZENG Zhaoyang, etal. Suppressing mislabeled data via grouping and self-attention[C]. The 16th European Conference on Computer Vision, Glasgow, UK, 2020: 786–802.
[18]
NORTHCUTT C G, JIANG Lu, and CHUANG I L. Confident learning: Estimating uncertainty in dataset labels[J]. Journal of Artificial Intelligence Research, 2021, 70: 1373–1411. doi: 10.1613/jair.1.12125
[19]
MA Xingjun, HUANG Hanxun, WANG Yisen, etal. Normalized loss functions for deep learning with noisy labels[C]. The 37th International Conference on Machine Learning, Vienna, Austria, 2020: 6543–6553.
[20]
ZHENG Songzhu, WU Pengxiang, GOSWAMI A, etal. Error-bounded correction of noisy labels[C]. The 37th International Conference on Machine Learning, Vienna, Austria, 2020: 11447–11457.
[21]
YI Kun and WU Jianxin. Probabilistic end-to-end noise correction for learning with noisy labels[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 7010–7018.
[22]
WANG Kai, PENG Xiaojiang, YANG Jianfei, etal. Suppressing uncertainties for large-scale facial expression recognition[C]. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, USA, 2020: 6896–6905.
[23]
ZENG Jiabei, SHAN Shiguang, and CHEN Xilin. Facial expression recognition with inconsistently annotated datasets[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 227–243.
[24]
WOO S, PARK J, LEE J Y, etal. CBAM: Convolutional block attention module[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 3–19.
[25]
HE Kaiming, ZHANG Xiangyu, REN Shaoqing, etal. Deep residual learning for image recognition[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 770–778.
[26]
HU Jie, SHEN Li, and SUN Gang. Squeeze-and-excitation networks[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018: 7132–7141.
[27]
KRIZHEVSKY A, SUTSKEVER I, and HINTON G E. ImageNet classification with deep convolutional neural networks[C]. The 25th International Conference on Neural Information Processing Systems, Lake Tahoe, USA, 2012: 1097–1105.
[28]
SIMONYAN K and ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[C]. The 3rd International Conference on Learning Representations, San Diego, USA, 2015: 1409–1556.
[29]
SANDLER M, HOWARD A, ZHU Menglong, etal. MobileNetV2: Inverted residuals and linear bottlenecks[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018: 4510–4520.
[30]
XIE Saining, GIRSHICK R, DOLLÁR P, etal. Aggregated residual transformations for deep neural networks[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 5987–5995.
[31]
HUANG Gao, LIU Zhuang, VAN DER MAATEN L, etal. Densely connected convolutional networks[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 2261–2269.
[32]
MA Ningning, ZHANG Xiangyu, ZHENG Haitao, etal. ShuffleNet V2: Practical guidelines for efficient CNN architecture design[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 122–138.
[33]
TAN Mingxing and LE Q V. EfficientNet: Rethinking model scaling for convolutional neural networks[C]. The 36th International Conference on Machine Learning, Long Beach, USA, 2019: 6105–6114.
LUO Yan, WANG Cong, YE Wenling. An Interpretable Prediction Model for Acute Kidney Injury Based on XGBoost and SHAP[J]. Journal of Electronics & Information Technology, 2022, 44(1): 27-38. doi: 10.11999/JEIT210931
LUO Yan, WANG Cong, YE Wenling. An Interpretable Prediction Model for Acute Kidney Injury Based on XGBoost and SHAP[J]. Journal of Electronics & Information Technology, 2022, 44(1): 27-38. doi: 10.11999/JEIT210931