Artificial Intelligence-Assisted ACR TI-RADS and Bethesda-like Risk Prediction for Indeterminate Thyroid Nodules: A Retrospective Proof-of-Concept Evaluation.
Authors
Affiliations (12)
Affiliations (12)
- Department of Radiology, Acıbadem Taksim Hospital, Istanbul 34373, Turkey.
- Department of Radiology, Faculty of Medicine, Gaziantep University, Gaziantep 27310, Turkey.
- Department of Radiology, Bucak State Hospital, Burdur 15300, Turkey.
- Department of Computer Engineering, Gedik University, Istanbul 34876, Turkey.
- Department of Otorhinolaryngology, School of Medicine, Biruni University, Istanbul 34295, Turkey.
- EPAM Systems Inc., Istanbul 34475, Turkey.
- Department of Radiology, Acibadem Kozyatagi Hospital, Istanbul 34734, Turkey.
- Department of General Surgery, Abdulkadir Yuksel State Hospital, Gaziantep 27100, Turkey.
- Department of Ear Nose and Throat Diseases Head and Neck Surgery, Medipol Acibadem Regional Hospital, Istanbul 34718, Turkey.
- Department of Radiology, Faculty of Medicine, Yeditepe University, Istanbul 34718, Turkey.
- Department of Radiology, American Hospital, Istanbul 34365, Turkey.
- Department of Nuclear Medicine, Faculty of Medicine, Gaziantep University, Gaziantep 27310, Turkey.
Abstract
<b>Background/Objectives</b>: Cytologically indeterminate thyroid nodules (Bethesda III-IV) remain a diagnostic challenge. We evaluated the TRAICK (version 1.9.4; TRAICK, Istanbul, Türkiye) artificial intelligence (AI) system, which generates an American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS) category and an imaging-derived Bethesda-like risk category from gray-scale ultrasound images. <b>Methods</b>: This retrospective proof-of-concept study included 52 surgically resected nodules with Bethesda III (<i>n</i> = 29) or IV (<i>n</i> = 23) cytology, with histopathology of the index nodule as the reference standard. The two TRAICK outputs were analyzed separately, and the ACR TI-RADS output was compared with two radiologists using the McNemar and DeLong tests. <b>Results</b>: Twenty-three nodules (44.2%) were malignant on index-nodule histopathology. The TRAICK ACR TI-RADS output had an accuracy of 71.2% (95% CI 57.7-81.7) and an area under the ROC curve (AUC) of 0.762 (0.640-0.884). Its AUC did not differ significantly from that of Radiologist 1 (0.762 vs. 0.777, <i>p</i> = 0.783); specificity was higher (65.5% vs. 37.9%, <i>p</i> = 0.021), whereas sensitivity was lower without a statistically significant difference (78.3% vs. 91.3%, <i>p</i> = 0.250). Compared with Radiologist 2, the TRAICK ACR TI-RADS output had higher accuracy (71.2% vs. 51.9%, <i>p</i> = 0.013), specificity (65.5% vs. 24.1%, <i>p</i> < 0.001), and AUC (0.762 vs. 0.629, <i>p</i> = 0.042). The Bethesda-like output had an AUC of 0.822 (0.707-0.938), with a sensitivity of 60.9% and specificity of 86.2% at the V-VI threshold. In sensitivity analyses, the AUC of the ACR TI-RADS output did not differ significantly from that of Radiologist 1, whereas its AUC advantage over Radiologist 2 was not statistically significant when low-risk neoplasms were classified as malignant. <b>Conclusions</b>: In this small, selected surgical cohort, the TRAICK ACR TI-RADS output showed discrimination that did not differ significantly from that of the more accurate of two experienced radiologists, with higher specificity, and its Bethesda-like output showed promising discrimination. These findings support further study of TRAICK as a complementary aid for preoperative decision-making; independent prospective validation is needed.