ChatGPT-4 Vision, an AI model for image interpretation, demonstrated diagnostic accuracy in evaluating thyroid nodules comparable to resident physicians, as shown in a study with 124 patients at Fujian Medical University. This AI model, analyzed using the C-TIRADS guidelines, achieved sensitivity of 86.2% and specificity of 60.0%, closely matching the resident's performance. While the attending physician performed better, the AI holds potential for supporting diagnostics in areas lacking expertise, despite limitations such as reliance on static images and a single-center study design.

Source: Frontiers