Abstract:
Objective The classification of parotid gland tumors is complex, qualitative analysis of dynamic enhancement scan is difficult to distinguish benign and malignant parotid tumors. To determine the clinical value of dynamic enhanced imaging in distinguishing the benign and malignant parotid gland mass through radiomics analysis of quantitative parameter images. Methods Magnetic resonance images of 51 cases of parotid gland mass confirmed by pathology in People's Hospital of Xinjiang Uygur Autonomous Region from January 2019 to April 2022 were retrospectively analyzed, including 12 pleomorphic adenomas, 19 Warthin tumors, 8 malignant tumors, and 15 non-neoplastic lesions. Quantitative parametric maps of transfer constant (Ktrans), extracellular volume fraction (Ve), plasma volume fraction (Vp), and outflow rate constant (Kep) were generated from dynamic enhancement images of parotid gland, and the imaging features were extracted by FAE software to establish a radiomics model for the diagnosis of benign and malignant parotid lesions. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and accuracy were used to evaluate the radiomics model and assess the efficacy of distinguishing benign and malignant parotid gland lesions. At the same time, pleomorphic adenoma, Warthin tumors and non-neoplastic lesions were separately compared with malignant parotid adenoma in terms of radiomics. Results Feature extraction and radiomics analysis of dynamic enhanced images were performed to determine the efficacy of benign and malignant parotid gland mass. The AUC, accuracy, sensitivity, and specificity were 0.612, 0.844, 0.500 and 0.875, respectively. When pleomorphic adenoma, Warthin's tumor and non-neoplastic lesions were separately compared with malignant parotid tumors, the AUC, accuracy, sensitivity, and specificity were 0.736, 0.781, 0.909, 0.714, and 0.886, 0.880, 0.933, 0.857, and 0.805, 0.781, 0.700, 0.818. Conclusion Radiomics analysis using dynamic enhanced quantitative images can preliminarily determine the benign and malignant parotid gland tumors. The better performance of radiomics is achieved in distinguishing different pathological subtypes of benign parotid tumors, non-neoplastic lesions and malignant parotid tumors.