Prediction of Benign and Malignant Thyroid Nodules in Ultrasound Images Based on
Convolutional Neural Network
WANG Hongjiea, ZHANG Endongb, YU Xiac
a. Department of Medical Equipment; b. Department of Otolaryngology, Head and Neck Surgery; c. Department of Ultrasound,
Weihai Maternal and Child Health Hospital, Weihai Shandong 264200, China
Abstract:Objective To explore the value of deep learning in the differential diagnosis of benign and malignant thyroid nodules.
Methods Convolutional neural network was used to collect and model the ultrasound thyroid nodules of 2659 anonymous and 480
patients. The benign and malignant thyroid nodules were predicted and the pathological results were used as the standard. Results
The positive rate, negative rate, sensitivity, and specificity of the sonographer were 84.3%, 90.5%, 95.1%, and 83.1%, respectively.
The positive rate, negative rate, sensitivity, and specificity of deep learning were 87.2%, 93.1%, 97.4%, and 87.3%, respectively.
Conclusion The convolutional neural network based on deep learning has high diagnostic sensitivity, diagnostic efficiency and
diagnostic specificity, which can effectively predict nodules in thyroid ultrasound images and can determine its benign and malignant.
王洪杰a,张恩东b,于霞c. 基于卷积神经网络的超声影像甲状腺结节良恶性预测研究[J]. 中国医疗设备, 2020, 35(1): 23-25.
WANG Hongjiea, ZHANG Endongb, YU Xiac. Prediction of Benign and Malignant Thyroid Nodules in Ultrasound Images Based on
Convolutional Neural Network. China Medical Devices, 2020, 35(1): 23-25.