TIRADS Scoring: Interpreting Your Thyroid Nodule Results

A TIRADS score is a rating system that radiologists assign to thyroid nodules based on how they look on ultrasound, and it tells you roughly how likely the nodule is to be cancerous. The most widely used version, ACR TI-RADS (from the American College of Radiology), sorts nodules into five tiers, from TR1 (benign) to TR5 (highly suspicious). In one study of over a thousand biopsied nodules, cancer rates were 0% for TR1 and TR2, about 0.3% for TR3, roughly 5% for TR4, and around 41% for TR5. If your report includes a TIRADS category, it is meant to guide the next step, whether that is reassurance, a follow-up ultrasound, or a biopsy, and the system was specifically designed to reduce unnecessary biopsies of nodules that are almost certainly harmless.

How the Score Is Calculated

ACR TI-RADS works by assigning points across five ultrasound features of the nodule: its composition (solid, partly cystic, or fully cystic), its echogenicity (how bright or dark it appears compared to surrounding tissue), its shape, its margins, and any echogenic foci (bright spots that may represent calcifications). Each feature has a range of possible descriptions, and each description carries a point value. A radiologist tallies the points, and the total determines the TIRADS category. A purely cystic nodule, for instance, scores zero across the board and lands at TR1. A solid, very dark nodule with irregular margins and tiny calcifications racks up points quickly and ends up at TR4 or TR5.

Not all features carry equal weight. Among the suspicious findings, a taller-than-wide shape (meaning the nodule grows more vertically than horizontally) stands out as a particularly strong predictor of cancer. One study found that this shape had an odds ratio of about 25 for malignancy, making it the single strongest independent marker among common ultrasound features.¹ Microcalcifications (tiny bright specks scattered within the nodule) and hypoechogenicity (a darker-than-normal appearance) also independently predict malignancy, though their odds ratios are considerably lower. The taller-than-wide feature is especially useful in small solid nodules under one centimeter, where it has shown high sensitivity and specificity for papillary thyroid cancer.²

Context matters, though. The suspicious features do not all behave the same way in every type of nodule. In solid hypoechoic nodules, any suspicious feature carries a high malignancy risk. But the same features in partly cystic nodules or in nodules that are the same brightness as surrounding tissue carry only intermediate risk.³ This is why the system adds up features rather than relying on any single one. A nodule with one suspicious characteristic in an otherwise reassuring background may only score a TR3 or low TR4.

What Each TIRADS Level Means for You

The five categories translate to escalating levels of concern, and each comes with a different recommendation for what happens next. Here is how they break down in practice:

TR1 (Benign): A completely cystic nodule or one that is entirely normal-looking. No biopsy needed, and usually no follow-up.
TR2 (Not suspicious): Includes classic benign patterns like simple cysts with a bright spot or spongiform nodules. No biopsy recommended.
TR3 (Mildly suspicious): Biopsy is generally recommended only if the nodule reaches 2.5 centimeters or larger. Otherwise, follow-up ultrasound monitoring.
TR4 (Moderately suspicious): Biopsy is typically recommended at 1.5 centimeters or larger.
TR5 (Highly suspicious): Biopsy is recommended when the nodule is 1 centimeter or larger.

These size thresholds are a critical and often misunderstood part of the system.⁴ A TR5 nodule that measures 7 millimeters does not automatically need a biopsy. The guidelines may call for surveillance instead. This point trips up a lot of patients who see “highly suspicious” on their report and assume surgery is imminent. The size cutoffs exist because very small cancers, even confirmed ones, often grow so slowly that immediate intervention may not be necessary. Your doctor should weigh the TIRADS category, the size, and your clinical picture together before deciding on next steps.

The real-world cancer rates behind these categories are reassuring at the lower end. In a large Canadian study, TR1 and TR2 nodules had a 0% cancer rate, TR3 had a 0.3% rate, TR4 had about a 5% rate, and TR5 reached roughly 41%.⁵ Even at TR5, more than half of nodules turn out to be benign. This is worth remembering if you have been told your nodule is “highly suspicious” and are panicking.

From Ultrasound Score to Biopsy Results

If your nodule meets the size and category threshold for biopsy, the next step is usually a fine-needle aspiration (FNA), where a thin needle extracts cells for examination under a microscope. The biopsy results come back classified under a different system called the Bethesda System for Reporting Thyroid Cytopathology, which grades the cells from Category I (nondiagnostic) through Category VI (malignant). Ideally, the TIRADS level and the Bethesda category should tell a consistent story, and in general they do. Research has found a statistically significant association between the two classification systems, with the strongest agreement between Bethesda IV (suspicious for a follicular neoplasm) and TR4.⁶

The negative predictive value of the system, meaning how reliably a low TIRADS score rules out cancer, is strong. In a large study comparing ACR TI-RADS scores with both FNA results and surgical pathology, TR3 nodules that were not recommended for biopsy turned out to be truly benign about 95% of the time.⁷ In other words, when the system says a nodule looks low-risk, it is usually right. The positive predictive value at the high end is more modest: even among TR5 nodules, the likelihood that a given nodule is actually malignant on histopathology was about two-thirds in that same study. The system is designed to cast a wide net at the top end, catching nearly all cancers while accepting that some benign nodules will look suspicious.

Why Different Radiologists Sometimes Disagree

One limitation you should be aware of is that TIRADS scoring involves human judgment, and two radiologists looking at the same ultrasound images may not always assign the same category. Studies examining interobserver agreement for ACR TI-RADS have found fair to moderate concordance for the overall score, with agreement on individual features like margins and shape being even lower.⁸ A blinded multicenter study found that the agreement between observers for the ACR TI-RADS category was moderate, with margins and shape being the features where radiologists disagreed most.⁹

The good news is that even when radiologists disagree on the specific score, they tend to agree on the decision that matters most: whether or not a biopsy is warranted. Research across multiple TIRADS systems has found that agreement on the biopsy recommendation is substantially higher than agreement on individual feature descriptions, and improves further after specific training.¹⁰ So even if two radiologists score your nodule slightly differently, they are likely to recommend the same course of action. If you find yourself with a borderline score and some uncertainty, requesting a second read or seeking an opinion at a high-volume thyroid center is reasonable.

Multiple TIRADS Systems Exist Around the World

ACR TI-RADS is the most commonly used system in North America, but it is not the only one. Europe has EU-TIRADS, South Korea has K-TIRADS, and several other national guidelines exist. These systems use similar ultrasound features but weight them differently and have different biopsy thresholds, which means the same nodule can get different recommendations depending on which system your radiologist follows.

A comparative study found that ACR TI-RADS had the highest diagnostic accuracy (about 63%) and the best specificity among the three most common systems, meaning it is the best at correctly identifying benign nodules and avoiding unnecessary biopsies. K-TIRADS had the highest sensitivity (about 94%), meaning it catches the most cancers but at the cost of recommending more biopsies overall.¹¹ A broader real-world comparison of six TIRADS systems found a similar pattern: ACR TI-RADS and the Chinese system (C-TIRADS) tended to recommend fewer biopsies, while K-TIRADS and the American Thyroid Association guidelines cast a wider net.¹²

The practical takeaway: if you are comparing notes with someone in another country who has a similar nodule but got a different recommendation, the guideline used may be the reason. Neither system is objectively “better” in all situations. ACR TI-RADS leans toward fewer interventions for low-risk nodules, which suits a healthcare environment concerned about overdiagnosis. K-TIRADS leans toward catching every possible cancer, even at the expense of more benign biopsies.

The Overdiagnosis Problem

Thyroid nodules are extraordinarily common. A global meta-analysis of over nine million individuals found the overall prevalence to be about 25%, and that figure climbs much higher when very sensitive ultrasound equipment is used, with one study using high-frequency probes detecting nodules in 68% of participants.¹³ ¹⁴ Despite this flood of detected nodules, thyroid cancer remains relatively rare.¹⁵ The gap between how often we find nodules and how often those nodules turn out to matter is the core of the overdiagnosis concern.

TIRADS was introduced partly to address this gap, and on that front, it has delivered measurable results. A meta-analysis of unnecessary biopsy rates found that ACR TI-RADS had the lowest rate at about 25%, compared to roughly 51% for ATA guidelines, 38% for EU-TIRADS, and 55% for K-TIRADS.¹⁶ Research on the ACR TI-RADS system specifically has found that it reduces the number of unnecessary biopsies of benign nodules by roughly 20 to 47% compared to other systems.¹⁷ Still, one audit found that after introducing TIRADS, the proportion of malignant nodules diagnosed by biopsy did not significantly change, though TR5 nodules were about five times more likely to be malignant than TR3 nodules.¹⁸ The system helps filter the flood, but it has not eliminated the problem entirely.

When Biopsy Results Are Unclear

Sometimes the biopsy itself does not give a clean answer. Bethesda categories III (atypia of undetermined significance) and IV (suspicious for a follicular neoplasm) are considered “indeterminate,” meaning the cells look unusual but are not clearly cancerous. Historically, many of these patients went straight to surgery to get a definitive answer, which meant some people had part or all of their thyroid removed for what turned out to be benign disease.

Molecular testing has become an increasingly common alternative. Tests like Afirma (a gene expression classifier) and ThyroSeq (a gene sequencing panel) analyze the biopsy sample at a molecular level to help determine whether the nodule is likely benign or malignant. A meta-analysis of these tools found that the newer generation tests, particularly ThyroSeq v3, performed well, with an area under the curve of 0.95.¹⁹ The strength of these tests varies by purpose. The newer Afirma GSC (Gene Sequencing Classifier) was particularly good at ruling out cancer (very few false negatives), while ThyroSeq v2 was better at ruling it in.²⁰ In practice, if your molecular test comes back “benign,” your doctor may feel comfortable recommending surveillance instead of surgery. If it comes back “suspicious,” surgery is more likely to be recommended.

TIRADS in Children

The standard ACR TI-RADS system was developed using adult data, and applying it directly to children creates problems. Thyroid nodules are less common in children, but when they do occur, a higher percentage turn out to be malignant compared to adults. Using adult TIRADS size thresholds in pediatric patients means some cancers get missed because the biopsy cutoffs are too generous for this population.

A multi-institutional review found that applying adult TI-RADS criteria would have recommended biopsy for only about 35% of pediatric nodules, versus roughly 77% under ATA pediatric guidelines. Using a modified approach where any score of TR3 or higher triggered a biopsy achieved 100% sensitivity, meaning no malignant nodules were missed, though specificity dropped to about 29%.²¹ A separate study examining pediatric-specific cutoffs proposed lowering the biopsy threshold to 0.5 centimeters for TR5 nodules and adjusting TR4 thresholds based on clinical risk factors like radiation exposure or family history, which improved sensitivity and reduced the rate of missed cancers.²² A smaller study of pediatric patients found that TR5 had a very high positive predictive value of about 88%, with all malignancies occurring in the TR4 and TR5 groups.²³

If your child has a thyroid nodule, the takeaway is that the standard adult TIRADS score on the report may underestimate the need for biopsy. Pediatric endocrinologists and surgeons who specialize in children often use lower thresholds or incorporate additional risk factors before making a biopsy decision.

Nodules Found by Accident on Other Scans

Many thyroid nodules are discovered not because someone was having thyroid problems, but because they showed up on a scan done for another reason. A CT of the neck for trauma, a PET/CT for cancer staging, or even a carotid ultrasound can reveal an unsuspected thyroid nodule. When a nodule lights up on a PET scan (meaning it is metabolically active), the malignancy rate is substantially higher than for nodules found on routine ultrasound, reported in the range of 20 to 40%.²⁴ Guidelines recommend that these incidentally detected nodules be followed up with a dedicated thyroid ultrasound, and TIRADS scoring can then be applied to decide whether a biopsy is warranted.²⁵ The usual size thresholds still apply: a TR4 or TR5 nodule found incidentally on PET should be biopsied if it meets the size criteria, just as it would if found on a routine thyroid ultrasound.

What Surveillance Looks Like in Practice

For many patients, the recommendation after a TIRADS assessment is not biopsy but follow-up imaging. This is sometimes called active surveillance, and it can feel uncomfortable, especially if you have been told your nodule is “mildly suspicious.” In practice, most nodules under surveillance stay stable. A study tracking over 550 nodules found that only about 2.5% showed a meaningful increase in diameter (3 mm or more) over time.²⁶ Interestingly, the same study found that volume measurements fluctuated much more than diameter measurements, with about a third of stable nodules showing volume swings of over 50%. This means that if your follow-up report mentions a volume change, it may not indicate real growth; volume is inherently noisier to measure. Diameter changes are considered more reliable for tracking genuine enlargement.

Typical follow-up intervals depend on the TIRADS category and the initial nodule size. A TR3 nodule under 2.5 centimeters might be re-imaged in one to two years, while a TR5 nodule under a centimeter that does not meet the biopsy threshold might be watched more closely with imaging every six to twelve months. If a nodule grows or develops new suspicious features, the recommendation may shift from surveillance to biopsy.

Artificial Intelligence and the Future of Scoring

One of the most active areas of research is using artificial intelligence to standardize TIRADS scoring and reduce the variability between radiologists. An AI model trained on ACR TI-RADS features achieved an area under the curve of 0.91 for distinguishing benign from malignant nodules, matching the performance of experienced radiologists and exceeding that of junior ones. Its specificity was actually higher than both groups, meaning it flagged fewer benign nodules as suspicious.²⁷ A larger multicenter study of a deep-learning system called ThyNet found that when radiologists used the AI as a second opinion, their diagnostic performance improved, and the number of biopsies recommended dropped from about 62% of nodules to 35% without increasing the rate of missed cancers.²⁸

These tools are not yet standard practice in most clinics, but they address one of the system’s real weaknesses. The interobserver variability that plagues human scoring could largely be eliminated by an algorithm that applies the same criteria to every image. As these AI tools get validated across more populations and built into ultrasound machines, they are likely to become part of routine thyroid imaging within the next several years.

The Economics of Getting It Right

The push toward risk stratification is not just about patient outcomes; it also has substantial financial implications. A cost-effectiveness analysis comparing ACR TI-RADS to the ATA 2015 guidelines found that TIRADS slightly dominated the ATA approach, producing marginally better health outcomes at lower cost, and was the more cost-effective option nearly 80% of the time in probabilistic modeling.²⁹ Strict adherence to ACR TI-RADS guidelines has been estimated to save regional healthcare systems meaningful amounts annually simply by eliminating biopsies that were never needed in the first place.³⁰ An Australian study found that the reduction in unnecessary biopsies under ACR TI-RADS was about 55%, translating to thousands of dollars in savings over the study period.³¹ The savings come not just from avoiding the biopsy procedure itself but from all the downstream costs of managing an unnecessary finding: repeat imaging, specialist consultations, patient anxiety, and occasional complications from procedures that were never clinically warranted.