Abstract: PURPOSE:This study was intended to use MaZda software to explore image texture analysis value of T1WI, T2WI, T1WI enhancement, DWI map in differential diagnosis of benign and malignant parotid gland tumors. METHODS: A total of 131 cases undergoing MRI examination and surgery during 2015 to 2019 were included. A radiologist used MaZda software to draw ROIs from the selected images which contained maximal layer of lesions to obtain individual texture feature values. B11 statistical software module from MaZda software was used to classify benign and malignant parotid gland tumors according to the selected texture features. RESULTS: The discriminant ability of texture analysis based on T2WI maps was better than that based on the other three sequence images. In raw data analysis(RDA), principal component analysis(PCA), linear discriminant analysis(LDA) and nonlinear discriminant analysis(NDA) of the four classification methods were carried out. the misclassification rate of NDA method to identify benign and malignant parotid gland tumor was significantly lower than other kinds of methods. CONCLUSION: Texture analysis based on T2WI maps can be used as an auxiliary method to provide reliable and objective evidence for differential diagnosis of benign and malignant parotid gland tumors. NDA is more effective than RDA,PCA,LDA in differential diagnosis. Attention should be paid to its generalization ability and sample-dependent characteristics in clinical application.

Key words: Benign and malignant parotid gland tumors, Magnetic resonance imaging, Texture analysis