CT-Based Radiomics Models for Differentiation of Benign and Malignant Thyroid Nodules: A Multicenter Development-and-Validation.

Background: CT is increasingly detecting thyroid nodules. Prior studies indicated a potential role of CT-based radiomics models in characterizing thyroid nodules, although lacked external validation. Objectives: To develop and validate a CT-based radiomics model for the differentiation of benign and malignant thyroid nodules. Methods: This retrospective study included 378 patients (mean age, 46.3±13.9 years; 86 men, 292 women) with 408 resected thyroid nodules (145 benign, 263 malignant) from two centers (center 1: 293 nodules, January 2018-December 2022; center 2: 115 nodules, January 2020-December 2022), who underwent preoperative multiphase neck CT (noncontrast, arterial, and venous phases). Nodules from center 1 were divided into training (n=206) and internal validation (n=87) sets; all nodules from center 2 formed an external validation set. Radiologists assessed nodules for morphologic CT features. Nodules were manually segmented on all phases, and radiomic features were extracted. Conventional (clinical and morphologic CT), noncontrast radiomics, arterial-phase radiomics, venous-phase radiomics, multiphase radiomics, and combined (clinical, morphologic, and multiphase radiomics) models were established using feature selection methods and evaluated by ROC curve analysis, calibration curves, and decision-curve analysis. Results: The combined model included patient age, three morphologic features (cystic change, edge interruption sign, abnormal cervical lymph nodes), and 28 radiomic features (from all three phases). In the external validation set, the combined model had AUC of 0.923 and, at an optimal threshold derived in the training set, sensitivity of 84.0%, specificity of 94.1%, and accuracy of 87.0%. In the external validation set, AUC was significantly higher for the combined model than for the conventional model (0.827), noncontrast radiomics model (0.847), arterial-phase radimoics model (0.826), venous-phase radiomics model (0.773), and multiphase radiomics model (0.824) (all p<.05). In the external validation set, the calibration curves indicated lowest (i.e., best) Brier score for the combined model; in decision-curve analysis, the combined model had the highest net benefit for most of the range of threshold probabilities. Conclusion: A combined model incorporating clinical, morphologic CT, and multiphasic radiomics CT features, exhibited robust performing in differentiating benign and malignant thyroid nodules. Clinical Impact: The combined radiomics model may help guide further management for thyroid nodules detected on CT.

Discrimination between HER2-overexpressing, -low-expressing, and -zero-expressing statuses in breast cancer using multiparametric MRI-based radiomics.

OBJECTIVES: To explore the performance of multiparametric MRI-based radiomics in discriminating different human epidermal growth factor receptor 2 (HER2) expressing statuses (i.e., HER2-overexpressing, HER2-low-expressing, and HER2-zero-expressing) in breast cancer. METHODS: A total of 771 breast cancer patients from two institutions were retrospectively studied. Five-hundred-eighty-one patients from Institution I were divided into a training dataset (n1 = 407) and an independent validation dataset (n1 = 174); 190 patients from Institution II formed the external validation dataset. All patients were categorized into HER2-overexpressing, HER2-low-expressing, and HER2-zero-expressing groups based on pathologic examination. Multiparametric (including T2-weighted imaging with fat suppression [T2WI-FS], diffusion-weighted imaging [DWI], apparent diffusion coefficient [ADC], and dynamic contrast-enhanced [DCE]) MRI-based radiomics features were extracted and then selected from the training dataset using the least absolute shrinkage and selection operator (LASSO) regression. Three predictive models to discriminate HER2-overexpressing vs. others, HER2-low expressing vs. others, and HER2-zero-expressing vs. others were developed based on the selected features. The model performance was evaluated using the area under the receiver operating characteristic curve (AUC). RESULTS: Eleven radiomics features from DWI, ADC, and DCE; one radiomics feature from DWI; and 17 radiomics features from DWI, ADC, and DCE were selected to build three predictive models, respectively. In training, independent validation, and external validation datasets, radiomics models achieved AUCs of 0.809, 0.737, and 0.725 in differentiating HER2-overexpressing from others; 0.779, 0.778, and 0.782 in differentiating HER2-low-expressing from others; and 0.889, 0.867, and 0.813 in differentiating HER2-zero-expressing from others, respectively. CONCLUSIONS: Multiparametric MRI-based radiomics model may preoperatively predict HER2 statuses in breast cancer patients. CLINICAL RELEVANCE STATEMENT: The MRI-based radiomics models could be used to noninvasively identify the new three-classification of HER2 expressing status in breast cancer, which is helpful to the decision-making for HER2-target therapies. KEY POINTS: • Detecting HER2-overexpressing, HER2-low-expressing, and HER2-zero-expressing status in breast cancer patients is crucial for determining candidates for anti-HER2 therapy. • Radiomics features from multiparametric MRI significantly differed among HER2-overexpressing, HER2-low expressing, and HER2-zero-expressing breast cancers. • Multiparametric MRI-based radiomics could preoperatively evaluate three different HER2-expressing statuses and help to determine potential candidates for anti-HER2 therapy in breast cancer patients.