ISSN 1470-3947 (print) | ISSN 1479-6848 (online)

Endocrine Abstracts (2019) 63 GP264 | DOI: 10.1530/endoabs.63.GP264

Assessment of different ti-rads systems using computer-aided detection technology

Kuen-Yuan Chen1, Ming-Hsun Wu1, Chiung-Nien Chen1 & Argon Chen2


1Department of Surgery, National Taiwan University Hospital, Taipei, Taiwan; 2College of Engineering, National Taiwan University, Taipei, Taiwan.


Aims: Different guidelines of Thyroid Imaging, Reporting, and Data System (TI-RADS) have been published to determine whether FNAC is required. Most of the guidelines recommended observation for nodules less than one centimeter and FNAC for nodule larger than two centimeters. In this study, we focus on the management of nodules size between one and two centimeters. We aim to evaluate the different guidelines on the treatment of thyroid nodules with the computer-aided detection (CADe) technology. We also propose a novel guideline for FNAC to reduce unnecessary procedures.

Methods and materials: The thyroid ultrasound pictures database collected prospectively were used. From July 2008 to December 2015, 369 patients with 422 nodules size between one and two centimeters were recruited, which were pathologically proven. (191 benign and 231 malignant) Eight guidelines, i.e., ACR (2017), ATA (2015), AACE (2016), Russ (2016), Seo (2016), BTA (2014), Kwak (2011) and Kwak (2013), were evaluated with sonographic features computerized by a USFDA-cleared CADe software device (AmCAD UT Detection). This CADe device quantifies and visualizes six sonographic features: hypoechogenicity, echogenic foci, heterogeneous texture, irregular margin, anechoic area, and taller-than-wide features, to assist physicians in making their diagnostic decisions with structural reports. The novel guidelines for the recommendation of FNAC were proposed using the quantified sonographic features.

Results: The ultrasound images were further divided into two groups, one is the size less 1.5 cm (1–1.5 cm) and the other is the size larger than 1.5 cm (1.5–2 cm). For group (1–1.5 cm), we presumed that the sensitivity above 90% was acceptable. The BTA (2014) had the sensitivity of 91%, with the specificity 23%. For group (1.5–2 cm), higher sensitivity is demanded. The Russ (2016) had the sensitivity of 96%, with the specificity 19%. The novel guideline was proposed and tested. For group (1–1.5 cm), if either one of US features including hypoechogenicity, calcifications and taller than wide was present, then sensitivity is 91% and specificity is 26%. For group (1.5–2 cm), if either one of US features including hypoechogenicity, calcifications and heterogeneity were present, the sensitivity is 95% and specificity is 21%. The guideline is simple and effective in reducing unnecessary FNAC procedures.

Conclusion: The CADe device is an effective tool to assist physicians in following the TI-RADS guidelines. With the proposed novel guideline, it’s shown that 91% of malignant nodules (1–1.5 cm) and 95% of malignant nodules (1.5–2) could be successfully identified and proceed to FNAC for further treatment with better specificity compared to other guidelines.

Article tools

My recent searches

No recent searches.