Search

Abstract Objective Capsular characteristics of pleomorphic adenoma (PA) has various forms. Patients without complete capsule has a higher risk of recurrence than patients with complete capsule. We aimed to develop and validate CT-based intratumoral and peritumoral radiomics models to make a differential diagnosis between parotid PA with and without complete capsule. Methods Data of 260 patients (166 patients with PA from institution 1 (training set) and 94 patients (test set) from institution 2) were retrospectively analyzed. Three Volume of interest (VOIs) were defined in the CT images of each patient: tumor volume of interest (VOItumor), VOIperitumor, and VOIintra-plus peritumor. Radiomics features were extracted from each VOI and used to train nine different machine learning algorithms. Model performance was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC). Results The results showed that the radiomics models based on features from VOIintra-plus peritumor achieved higher AUCs compared to models based on features from VOItumor. The best performing model was Linear discriminant analysis, which achieved an AUC of 0.86 in the tenfold cross-validation and 0.869 in the test set. The model was based on 15 features, including shape-based features and texture features. Conclusions We demonstrated the feasibility of combining artificial intelligence with CT-based peritumoral radiomics features can be used to accurately predict capsular characteristics of parotid PA. This may assist in clinical decision-making by preoperative identification of capsular characteristics of parotid PA.

ObjectiveThis study aims to investigate the value of machine learning models based on clinical-radiological features and multiphasic CT radiomics features in the differentiation of benign parotid tumors (BPTs) and malignant parotid tumors (MPTs).MethodsThis retrospective study included 312 patients (205 cases of BPTs and 107 cases of MPTs) who underwent multiphasic enhanced CT examinations, which were randomly divided into training (N = 218) and test (N = 94) sets. The radiomics features were extracted from the plain, arterial, and venous phases. The synthetic minority oversampling technique was used to balance minority class samples in the training set. Feature selection methods were done using the least absolute shrinkage and selection operator (LASSO), mutual information (MI), and recursive feature extraction (RFE). Two machine learning classifiers, support vector machine (SVM), and logistic regression (LR), were then combined in pairs with three feature selection methods to build different radiomics models. Meanwhile, the prediction performances of different radiomics models based on single phase (plain, arterial, and venous phase) and multiphase (three-phase combination) were compared to determine which model construction method and phase were more discriminative. In addition, clinical models based on clinical-radiological features and combined models integrating radiomics features and clinical-radiological features were established. The prediction performances of the different models were evaluated by the area under the receiver operating characteristic (ROC) curve (AUC) and the drawing of calibration curves.ResultsAmong the 24 established radiomics models composed of four different phases, three feature selection methods, and two machine learning classifiers, the LASSO-SVM model based on a three-phase combination had the optimal prediction performance with AUC (0.936 [95% CI = 0.866, 0.976]), sensitivity (0.78), specificity (0.90), and accuracy (0.86) in the test set, and its prediction performance was significantly better than with the clinical model based on LR (AUC = 0.781, p = 0.012). In the test set, the combined model based on LR had a lower AUC than the optimal radiomics model (AUC = 0.933 vs. 0.936), but no statistically significant difference (p = 0.888).ConclusionMultiphasic CT-based radiomics analysis showed a machine learning model based on clinical-radiological features and radiomics features has the potential to provide a valuable tool for discriminating benign from malignant parotid tumors.

Background: Nasopharyngeal carcinoma (NPC) is a common tumor in China. Accurate stages of NPC are crucial for treatment. We therefore aim to develop radiomics models for discriminating early-stage (I–II) and advanced-stage (III–IVa) NPC based on MR images. Methods: 329 NPC patients were enrolled and randomly divided into a training cohort (n = 229) and a validation cohort (n = 100). Features were extracted based on axial contrast-enhanced T1-weighted images (CE-T1WI), T1WI, and T2-weighted images (T2WI). Least absolute shrinkage and selection operator (LASSO) was used to build radiomics signatures. Seven radiomics models were constructed with logistic regression. The AUC value was used to assess classification performance. The DeLong test was used to compare the AUCs of different radiomics models and visual assessment. Results: Models A, B, C, D, E, F, and G were constructed with 13, 9, 7, 9, 10, 7, and 6 features, respectively. All radiomics models showed better classification performance than that of visual assessment. Model A (CE-T1WI + T1WI + T2WI) showed the best classification performance (AUC: 0.847) in the training cohort. CE-T1WI showed the greatest significance for staging NPC. Conclusion: Radiomics models can effectively distinguish early-stage from advanced-stage NPC patients, and Model A (CE-T1WI + T1WI + T2WI) showed the best classification performance.