The application of machine learning models based on ultrasonographic radiomics in the differentiation of benign and malignant salivary gland tumors
ZHANG Shunchu
XIA Zhen
MIAO Qing
HUANG Xiaochen
ZHANG Wei
Abstract：Objective To investigate the diagnostic value of ultrasound radiomic-based machine learning models in differen-tiating between benign and malignant parotid gland tumors.Methods A total of 144 patients with solitary parotid gland tumors confirmed by pathology were selected,among whom 96 were benign(benign group)and 48 were malignant(malignant group).All patients underwent grayscale ultrasound examinations.Regions of interest(ROI)were manually delineated by an experi-enced ultrasound physician,and radiomic features were extracted using the Python-based Pyradiomics package.Using pathologi-cal results as the reference standard,the radiomic features with the highest discriminative value for differentiating between be-nign and malignant tumors were selected.Based on the selected features,multiple machine learning models were constructed,and their diagnostic performance was evaluated using receiver operating characteristic curves and the corresponding area under the curve(AUC).Results A total of six radiomic features were selected for model construction.All the models demonstrated good performance in the training set.Among them,the K-nearest Neighbor(KNN)model had the highest AUC in the validation set(0.88),followed by the Support Vector Machine(SVM)model(AUC=0.86).Conclusion Ultrasound radiomic-based ma-chine learning models show favorable diagnostic efficacy in differentiating between benign and malignant parotid gland tumors.
Keywords：Parotid gland tumorsUltrasoundRadiomicsMachine learning
Publication Date：2025-08-30
Online Publishing Date：2025-09-28(First online date of this platform, not the publication date of the document)
Pages：4( 24-27 )