Efficiency comparison of Kwak and ACR ( 2017 ) Thyroid Imaging Reporting and Data System ( TI‐RADS) classification :a polycentric retrospective study / 中华超声影像学杂志

ABSTRACT

Objective To evaluate the diagnostic efficacy of Kwak and ACR( 2017 ) thyroid imaging reporting and data systems ( T I‐RADS ) for thyroid nodules . Methods Cases of thyroid nodule who underwent surgery from January 2015 to M arch 2018 in 15 hospitals in Sichuan province were collected and the ultrasonographic features of thyroid nodules were retrospectively analyzed by trained senior ultrasound physicians using Kwak and ACR T I‐RADS classification methods . Totally ,12 712 thyroid nodules were observed ,7 023 thyroid nodules in 7 023 cases with complete ultrasound and surgical and pathological data were eventually enrolled in the study . T hyroid nodules with solid ,hypoechoic or very hypoechoic ,tall/wide ratio ≥ 1 , margin ill‐defined and microcalcification were classified as malignant signs of ultrasound . M alignant percentage was calculated and diagnostic tests were performed . Results ① T here was a statistical difference between the benign and malignant nodules in the two types of T I‐RADS classification ( P＜0 .01) . ② T he area under ROC curve of Kwak and ACR in the diagnosis of malignant nodules were 0 .89 and 0 .84 ,respectively . T he Youden index of Kwak and ACR were 0 .66 and 0 .57 ,respectively . ③Taking Kwak T I4B and ACR T R4 as critical points for malignancy ,the sensitivity ,specificity ,positive predictive value and negative predictive value of Kwak T I 4B were 75 .0% ,90 .9% ,83 .2% ,and 85 .9% , respectively . T he accuracy of Kwak T I4B was 84 .9% ; T he sensitivity ,specificity ,positive predictive value and negative predictive value of ACR T R4 were 88 .2% ,68 .9% ,62 .9% ,and 90 .8% ,respectively . T he accuracy of ACR T R4 was 76 .2% . T he Kappa value of Kwak TI4B and ACR T R4 was 0 .52 . T he χ2 value of Kwak T I4B and ACR T R4 was 2 174 .6 ( P ＜ 0 .01 ) . Conclusions T he diagnostic values of two T I‐RADS classification methods for thyroid malignant nodules are high . T he overall efficiency of Kwak T I‐RADS classification method is better than that of ACR TI‐RADS classification method .