Value of Whole-Thyroid CT-Based Radiomics in Predicting Benign and Malignant Thyroid Nodules

Abstract

The objective of this research is to explore the value of whole-thyroid CT-based radiomics in predicting benign (noncancerous) and malignant thyroid nodules. The imaging and clinical data of 161 patients with thyroid nodules that were confirmed by pathology were retrospectively analyzed. The entire thyroid regions of interest (ROIs) were manually sketched for all 161 cases. After extracting CT radiomic features, the patients were divided into a training group (128 cases) and a test group (33 cases) according to the 4:1 ratio with stratified random sampling (fivefold cross validation). All the data were normalized by the maximum absolute value and screened using selection operator regression analysis and K best. The data generation model was trained by logistic regression. The effectiveness of the model in differentiating between benign and malignant thyroid nodules was validated by a receiver operating characteristic (ROC) curve. After data grouping, eigenvalue screening, and data training, the logistic regression model with the maximum absolute value normalized was constructed. For the training group, the area under the ROC curve (AUC) was 94.4% (95% confidence interval: 0.941–0.977); the sensitivity and specificity were 89.7% and 86.7%, respectively; and the diagnostic accuracy was 87.6%. For the test group, the AUC was 94.2% (95% confidence interval: 0.881–0.999); the sensitivity and specificity were 89.4% and 86.8%, respectively; and the diagnostic accuracy was 87.6%. The CT radiomic model of the entire thyroid gland is highly efficient in differentiating between benign and malignant thyroid nodules.

A Novel Interpretable Radiomics Model to Distinguish Nodular Goiter From Malignant Thyroid Nodules

“…The findings of the study demonstrate that a machine learning model using quantitative radiomics features can accurately predict the malignancy of TNs. In contrast to previous studies, 9,[27][28][29] we incorporated multimodel fusion during model construction and implemented model interpretation methods to enhance decision transparency. The fusion model's radiomics signature displayed a higher predictive value compared with using only a single LASSO regression model.…”

“…However, many thyroid diseases have multiple nodules, irregular shapes, and ambiguous boundaries, making it difficult to accurately describe thyroid lesions. In addition, individual nodules cannot reflect the full picture of thyroid lesions 9 …”

“…In addition, individual nodules cannot reflect the full picture of thyroid lesions. 9 Machine learning is increasingly being introduced to construct computer-aided diagnosis systems because of its strong ability to extract complex relationships from biomedical data. 10 Commonly used machine learning techniques include K-nearest neighbor, support vector machine (SVM), and random forest (RF).…”

Diagnosis of benign and malignant thyroid nodules by a dual-layer spectral detector CT-based nomogram

“…Similarly, DECT has improved the noninvasive evaluation of thyroid nodules, and an increasing number of studies have indicated that quantitative parameters derived from DECT, especially quantitative IC, can distinguish benign from malignant thyroid nodules (6,9,(14)(15)(16). However, there are few relevant studies exploring the applications of DECT, and some of them show contradictory results.…”

Ultrasound radiomics based XGBoost model to differential diagnosis thyroid nodules and unnecessary biopsy rate: Individual application of SHapley additive exPlanations

“…In recent years, ultrasound-based radiomics has been applied to the prediction of cervical lymph node metastasis [7][8][9] as well as BRAF-V600E mutation in thyroid cancer. 10,11 Meanwhile, this technique has been widely used in discriminating the nature of thyroid nodules [12][13][14][15] and achieved good results. However, its fusion models with artificial intelligence (AI) are diverse, and its diagnostic performance is also uneven.…”