A Clinical Decision Support System Using Ultrasound Textures and Radiologic Features to Distinguish Metastasis From Tumor-Free Cervical Lymph Nodes in Patients With Papillary Thyroid Carcinoma.

摘要: Objectives This study investigated the potential of a clinical decision support approach for classification metastatic and tumor-free cervical lymph nodes (LNs) in papillary thyroid carcinoma on basis radiologic textural analysis through ultrasound (US) imaging. Methods In this research, 170 LNs were examined by proposed method. To discover difference between groups, US imaging was used extraction features. The features B-mode scans included echogenicity, margin, shape, presence microcalcification. extract features, wavelet transform applied. A vector machine classifier to classify LNs. Results training set data, combination represented best performance with sensitivity, specificity, accuracy, area under curve (AUC) values 97.14%, 98.57%, 97.86%, 0.994, respectively, whereas based alone yielded lower performance, AUCs 0.964 0.922. On testing data set, model could an AUC 0.952, which corresponded accuracy 93.33%, 96.66%, 95.00%. Conclusions method has characterize via 2-dimensional US. Therefore, it can be as supplementary technique daily practice improve radiologists' understanding conventional characterizing