A generally trained artificial intelligence (AI) automatic diagnosis system can differentially diagnose benign and rare malignant thyroid carcinoma nodules, such as those of medullary thyroid carcinoma (MTC), with high accuracy, according to a new study published in Endocrine

“It may assist radiologists for screening of rare malignancy nodules that even senior radiologists are not acquainted with,” the authors of the study wrote.

A team of researchers led by Dexing Kong set out to assess the application value of such a system in the diagnosis of rare thyroid carcinomas like follicular, medullary, and anaplastic thyroid carcinomas as well as primary thyroid lymphoma, and to compare its performance with that of radiologists with different levels of experience. They retrospectively analyzed 342 patients who had a total of 378 thyroid nodules. Of these, 196 were rare malignant nodules. 

Continue Reading

Read more about the diagnosis of MTC

The team used postoperative pathology as the gold standard. They compared the diagnostic performances of 3 radiologists with different levels of expertise and that of the AI automatic diagnosis system.

The results showed that the AI system had an accuracy of 0.825 in diagnosing malignancy. This was higher than that of the radiologists—regardless of their level of expertise. 

The diagnostic sensitivities of the mid-level and senior radiologists were higher than that of the AI automatic diagnosis system. However, their specificities were much lower than that of the automatic system, at 0.533 and 0.478, respectively, compared to 0.802. 

The sensitivity and specificity of the junior radiologist were relatively balanced, but they were both lower than those of the AI automatic diagnosis system.

“In conclusion, our study showed that the AI automatic diagnosis system exhibited high diagnostic accuracy in the malignancy diagnosis of rare malignancy thyroid carcinomas in presence of solid benign nodules,” the researchers wrote. “The AI automatic diagnosis system can be potentially used as an auxiliary method for distinguishing rare malignancy thyroid nodules.”


Wang Y, Xu L, Lu W, et al. Clinical evaluation of malignancy diagnosis of rare thyroid carcinomas by an artificial intelligent automatic diagnosis system. Endocrine. Published online December 3, 2022. doi:10.1007/s12020-022-03269-4