Application of Artificial Intelligence Assistant Diagnosis System Based on C-TIRADS in Diagnosis of Benign and Malignant Suspicious Thyroid Nodules / 中国医学影像学杂志

ABSTRACT

Purpose To explore the application value of artificial intelligence(AI)assistant diagnosis system based on Chinese thyroid imaging reporting and data system(C-TIRADS)in the diagnosis of suspected thyroid nodules.Materials and Methods A total of 441 patients(445 nodules)with suspected thyroid nodules examined in Huadong Sanatorium from January 2018 to December 2021 were retrospectively analyzed.The low seniority group,the senior seniority group and the AI group were composed of the junior doctors,the senior doctors and the AI assistant diagnosis system,respectively.The suspicious thyroid nodules were classified according to the C-TIRADS guideline,and then the suspicious thyroid nodules were classified again by the low seniority group and the senior group combined with the AI assistant diagnosis system.Taking the pathological results as the gold standard,the diagnostic efficacy of each group was compared,and the sensitivity,specificity,accuracy,positive predictive value,negative predictive value,Jordan index and area under the curve(AUC)were further compared.Results Of the 445 thyroid nodules,there were 277 malignant nodules and 168 benign nodules.The AUCs of thyroid nodules in low seniority group,high seniority group and AI group were 0.760,0.858 and 0.849,respectively.The AUCs in high seniority group and AI group were significantly higher than that in low seniority group(both P<0.001),and the AUC in high seniority group was similar to that in AI group(P>0.05).The AUCs of thyroid nodules in low seniority group and high seniority group combined with AI assistant diagnosis system was higher than those in low seniority group and high seniority group,especially in low seniority group(0.830 vs.0.760,P<0.001).The accuracy,Jordan index and AUC were the highest when TR4C was used as the cut-off point of suspected thyroid nodules.Conclusion The AI assistant diagnosis system based on C-TIRADS has high value in the differential diagnosis of thyroid suspicious nodules,and the combined model can improve the diagnostic efficiency of different seniority doctors.The best cut-off point of different seniority doctors and combined model of AI assistant diagnosis system for differential diagnosis of thyroid suspicious nodules may be TR4C.