Comparison of six machine learning methods for differentiating benign and malignant thyroid nodules using ultrasonographic characteristics

BMC Medical Imaging

Research

Xuehao Gong¹  Leidan Huang²  Xiaogang Li³  Xianfen Diao³  Weixiang Liu³  Tiantian Pang⁴  Jianguang Liang⁵ 

[1] Department of Ultrasound, First Affiliated Hospital of Shenzhen University, Second People’s Hospital of Shenzhen, 518035, Shenzhen, China;Guangzhou Medical University, 510182, Guangzhou, China;Department of Ultrasound, First Affiliated Hospital of Shenzhen University, Second People’s Hospital of Shenzhen, 518035, Shenzhen, China;Health Science Center, Shenzhen University, 518060, Shenzhen, China;School of Biomedical Engineering, Shenzhen University, 518060, Shenzhen, China;Guangdong Key Laboratory for Biomedical Measurements and Ultrasound Imaging, 518060, Shenzhen, China;National-Regional Key Technology Engineering Laboratory for Medical Ultrasound, 518060, Shenzhen, China;Health Science Center, Shenzhen University, 518060, Shenzhen, China;School of Biomedical Engineering, Shenzhen University, 518060, Shenzhen, China;Guangdong Key Laboratory for Biomedical Measurements and Ultrasound Imaging, 518060, Shenzhen, China;National-Regional Key Technology Engineering Laboratory for Medical Ultrasound, 518060, Shenzhen, China;College of Computer Science and Technology, Jilin University, 130012, Changchun, China;School of Pharmacy & School of Biological and Food Engineering, Changzhou University, 213164, Changzhou, Jiangsu, China;

关键词: Machine learning;    Support vector machine;    Logistic regression;    Linear discriminant analysis;    Random forest;    GlmNet;    K-nearest neighbors;    Thyroid nodule;    Paired t-test;   

DOI  :  10.1186/s12880-023-01117-z

 received in 2022-12-25, accepted in 2023-10-02,  发布年份 2023

来源: Springer

【 摘 要 】

BackgroundSeveral machine learning (ML) classifiers for thyroid nodule diagnosis have been compared in terms of their accuracy, sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and area under the receiver operating curve (AUC). A total of 525 patients with thyroid nodules (malignant, n = 228; benign, n = 297) underwent conventional ultrasonography, strain elastography, and contrast-enhanced ultrasound. Six algorithms were compared: support vector machine (SVM), linear discriminant analysis (LDA), random forest (RF), logistic regression (LG), GlmNet, and K-nearest neighbors (K-NN). The diagnostic performances of the 13 suspicious sonographic features for discriminating benign and malignant thyroid nodules were assessed using different ML algorithms. To compare these algorithms, a 10-fold cross-validation paired t-test was applied to the algorithm performance differences.ResultsThe logistic regression algorithm had better diagnostic performance than the other ML algorithms. However, it was only slightly higher than those of GlmNet, LDA, and RF. The accuracy, sensitivity, specificity, NPV, PPV, and AUC obtained by running logistic regression were 86.48%, 83.33%, 88.89%, 87.42%, 85.20%, and 92.84%, respectively.ConclusionsThe experimental results indicate that GlmNet, SVM, LDA, LG, K-NN, and RF exhibit slight differences in classification performance.

【 授权许可】

CC BY   
© BioMed Central Ltd., part of Springer Nature 2023

【 图 表 】

Fig. 1

Fig. 3