AI Can Improve Thyroid Nodule Ultrasound Interpretations

Background: Accurate evaluation of thyroid nodules by ultrasound is critical to determining appropriate patient management. Guidelines have been developed to stratify the potential for malignancy using specific ultrasound findings, which has improved this approach. However, maintaining consistency of these interpretations remains challenging due to multiple factors. Objective: To determine whether artificial intelligence (AI) techniques can produce accurate interpretations of thyroid nodule ultrasound examinations. Design: Literature search. Methods: Journal articles that evaluated the use of AI methods to analyze the results of thyroid nodule ultrasound examinations that had cytologic and/or histologic correlation were reviewed. Algorithms used by AI methods analyzed a variety of the thyroid nodule ultrasound findings. Results: Of 1257 articles reviewed, only 27 studies met the inclusion criteria. Most studies used B-mode ultrasound techniques. Radiomic data were used by the AI tech more...