Thyroid Imaging Reporting and Data System
ACR TI-RADS is a standardized way to describe a thyroid nodule on ultrasound and translate its appearance into an estimate of cancer risk. The radiologist scores several imaging features, adds up the points, and lands the nodule in one of five risk levels, which — together with nodule size — guides whether to recommend a needle biopsy, follow-up scanning, or no action. Its central goal is to reduce unnecessary biopsies of nodules very unlikely to be cancer while still catching those that warrant attention. Note that 'TI-RADS' is best understood as a family: regional European, Korean, and Chinese versions are also widely used.
The first and current ACR-sanctioned system. It introduced a points-based scoring model across five standardized feature categories, designed to be more reproducible and biopsy-sparing than earlier ad hoc thyroid risk schemes.
Before the ACR effort, the 'TI-RADS' idea was proposed and iterated by individual investigators (e.g. Horvath in 2009, Kwak in 2011). These laid conceptual groundwork but were not ACR products and used different scoring logic; the 2017 ACR system is a distinct, committee-developed standard.
ACR TI-RADS is a points-based scheme: the reader assesses the nodule across five feature axes — composition, echogenicity, shape, margin, and echogenic foci — assigns points for each, and sums them. The total maps the nodule to one of five categories, TR1 through TR5, running from benign to highly suspicious. The recommendation to biopsy or to follow with repeat ultrasound then depends on both the TR level and the nodule's maximum size. Read the exact point values, size thresholds, and management cut-offs from the official ACR source.
These are our plain-language summaries. For the exact criteria, thresholds, and management rules, see the official source.
An interpretable radiomics framework mapping quantitative US features to TI-RADS semantics achieved ROC-AUC 0.941 ± 0.004 for benign vs. malignant thyroid nodule classification across 5,542 multicenter nodules — with SHAP analysis confirming texture heterogeneity as the dominant…
C-TIRADS ≥4B yielded 75.7% sensitivity, 75.8% specificity (AUC 0.784) for thyroid malignancy—not significantly different from Kwak TI-RADS (P>0.05) in 272 surgically confirmed cases (retrospective).
ACR TI-RADS and K-TIRADS show substantial inter-system agreement (κ=0.61) and similar AUC (p=0.52) across 4 readers, but ACR TI-RADS yields higher sensitivity and NPV at the cost of lower specificity and more unnecessary biopsies — retrospective, single cohort.
TI-RADS is best thought of as a family. The ACR version is points-based, summing weighted features into TR1–TR5. EU-TIRADS (European Thyroid Association, 2017) and K-TIRADS (Korean Society of Thyroid Radiology, 2016, revised 2021) are instead pattern-based: the reader matches a nodule to a prototypical 'look' and reads off a five-level category. C-TIRADS (China, 2020) takes a feature-count approach — adding a point per suspicious feature and subtracting one for a reassuring comet-tail artifact, then mapping the tally to a category. Head-to-head studies generally find broadly comparable discriminatory accuracy across systems, with meaningful differences in how aggressively each triggers biopsy (ACR tends to be the most biopsy-sparing; the 2016 K-TIRADS was among the most aggressive, which the 2021 revision narrowed).
Across validation cohorts, ACR TI-RADS reliably stratifies malignancy risk and is consistently the most effective major system at cutting unnecessary biopsies of benign nodules. A network meta-analysis comparing several ultrasound risk-stratification systems found generally similar overall diagnostic performance with trade-offs in sensitivity versus specificity, and other work reports higher specificity but lower sensitivity for ACR TI-RADS relative to EU- and 2016 K-TIRADS, alongside markedly lower unnecessary-biopsy rates. The system's main documented weakness is interobserver variability — margin and echogenic-foci assessment are the least reproducible features — which has motivated interest in AI-assisted scoring.
RadPigeon is an independent radiology news digest and is not affiliated with or endorsed by American College of Radiology (ACR) — for ACR TI-RADS; regional variants exist (see comparison). “TI-RADS” is a trademark of its owner and is named here only to refer to the system. Always consult the official source for the exact, current criteria.
