The Feasibility of Classification of Thyroid Nodules Integrated Expriences Based Inference of Radiologist and Extracted Feature Vectors in Ultrasound Images

Thyroid nodules are exceedingly common, and most nodules are benign. The classification of thyroid nodules can greatly reduce the cost of surgery. Current research focuses on the classifications of benign and malignant tumors in the diagnosis of thyroid nodules, but there are few studies that are based on the combination of CAD and the expriences based inference of radiologists. Following the necessity of manual intervention in diagnoses that use a CAD system, we prospectively developed a new method for classifying thyroid nodules that integrate human inference and classifiers for ultrasound images.