Artificial intelligence and elastography in diagnostic work-up of thyroid nodules: a systematic review and meta-analysis.

BACKGROUND

Artificial intelligence (AI) characterizes thyroid nodules by automatically extracting features from ultrasound images, whereas elastography quantitatively assesses tissue stiffness to aid in discriminating between benign and malignant cases. This study systematically evaluated the diagnostic accuracy of combining AI with elastography in differentiating benign and malignant thyroid nodule.

METHODS

A comprehensive literature search was conducted in the databases of PubMed, Embase, Web of Science, and the China National Knowledge Infrastructure (CNKI) using a predefined "subject + keyword" strategy to locate diagnostic studies on the combined use of elastography and AI in differentiating benign and malignant thyroid nodule. Diagnostic test performance was evaluated by generating summary receiver operating characteristic (SROC) curves and calculating pooled estimates of sensitivity and specificity. The methodological quality of the included studies was appraised using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. Data analysis was performed using Review Manager 5.4 and Stata 17.

RESULTS

A total of 19 studies comprising 4,655 ultrasound images of thyroid nodule were included. Compared with available technologies, the combination of AI and elastography demonstrated superior diagnostic performance for thyroid diseases, with a pooled sensitivity of 0.88 [95% confidence interval (CI): 0.84-0.91], a specificity of 0.91 (95% CI: 0.88-0.94), a diagnostic odds ratio (DOR) of 73.89 (95% CI: 40.49-134.85), and an area under the curve (AUC) of 0.95 (95% CI: 0.93-0.97).

CONCLUSIONS

AI combined with elastography has high diagnostic accuracy for thyroid nodule, and this promising technology is expected to be integrated into routine clinical practice to improve the diagnosis and prognosis of thyroid nodule.