How TI-RADS and the Bethesda System Decide Which Thyroid Nodules Get a Biopsy

What these two systems actually measure

TI-RADS and the Bethesda System do not diagnose thyroid cancer. Each converts a snapshot, one an ultrasound pattern, the other a needle sample of cells, into a graded probability of malignancy. Those probabilities, paired with nodule size, set the threshold for who gets a biopsy and who is watched. Reading their outputs correctly means treating a category as a risk estimate, not a verdict.

Thyroid nodules are extremely common, and the large majority are benign. The clinical problem is rarely finding nodules; it is deciding which of the many deserve a needle. Both systems exist to answer that triage question with structured numbers rather than intuition.

How ACR TI-RADS scores an ultrasound

The American College of Radiology TI-RADS, set out in the 2017 committee white paper published in the Journal of the American College of Radiology, scores a nodule across five ultrasound feature categories: composition, echogenicity, shape, margin, and echogenic foci. Each feature earns points, with more suspicious appearances earning more. A purely cystic nodule earns almost nothing; a solid, very hypoechoic nodule that is taller than wide, with irregular margins and punctate echogenic foci, accumulates a high total.

The summed points map to a level from TR1 (benign) to TR5 (highly suspicious). The key design choice is what happens next. TI-RADS pairs the level with the nodule's largest diameter to produce one of three actions: recommend fine-needle aspiration, recommend interval ultrasound follow-up, or do neither. A TR5 nodule crosses the biopsy threshold at a small size, while a mildly suspicious TR3 nodule is left alone until it is considerably larger. The system deliberately tolerates leaving small, low-suspicion nodules unbiopsied, because the pretest probability of clinically meaningful cancer in that group is low and the harms of chasing every one are real.

What TI-RADS measures is pattern, not tissue. It estimates the chance that a nodule with a given appearance is malignant, and it is explicit that its thresholds are calibrated to reduce unnecessary biopsies rather than to catch every cancer regardless of cost.

How the Bethesda System scores the cells

If a nodule crosses the biopsy threshold, a needle draws out cells, and a cytopathologist reports the result using the Bethesda System for Reporting Thyroid Cytopathology. Its third edition, published in the journal Thyroid in 2023, keeps six categories: nondiagnostic, benign, atypia of undetermined significance (AUS), follicular neoplasm, suspicious for malignancy, and malignant.

Each category carries an implied risk of malignancy drawn from pooled outcome data. Benign cytology carries a low single-digit risk. The two indeterminate categories in the middle, AUS and follicular neoplasm, are where the system is most honest about uncertainty: their malignancy risks sit in the rough range of the teens to the low thirties, which is precisely why they cannot be treated as either reassurance or diagnosis. Suspicious-for-malignancy and malignant categories carry high risks, the malignant category approaching near-certainty. The 2023 edition also refined the indeterminate zone, subdividing AUS to reflect differences in risk and the growing role of molecular testing.

Bethesda measures cells, not appearance. It estimates the chance that the sampled tissue is cancer, and like TI-RADS it reports a probability band rather than a yes-or-no answer.

Where the two systems meet

The systems are sequential, and their relationship is the interesting part. A 2023 concordance study published in JAMA Network Open compared ACR TI-RADS classification against Bethesda cytology and, in resected nodules, against final histopathology. The direction of the findings is what matters here. Lower TI-RADS scores carried strong negative predictive value, meaning a reassuring ultrasound pattern was usually confirmed benign. Higher TI-RADS categories tracked with meaningfully higher malignancy rates on final pathology, and the largest, most suspicious nodules carried the highest risk.

That concordance is a validation of the triage logic, not proof that imaging can replace a needle. The same study, like the guidelines themselves, points toward refining which nodules are aspirated by combining size and score, rather than biopsying on appearance alone. Two imperfect estimators, applied in sequence, filter a large population of mostly benign nodules down to the smaller group where surgery or close surveillance is genuinely warranted.

Why thresholds, not certainty, are the point

Both systems accept that some cancers will be watched rather than immediately sampled, because many small, low-risk thyroid cancers grow slowly and aggressive biopsy of every nodule generates its own harms: bleeding, anxiety, indeterminate results that trigger more procedures, and overtreatment of cancers that would never have caused symptoms. The size-and-score thresholds are the mechanism by which the guidelines trade a small amount of sensitivity for a large reduction in low-value biopsies. Understanding that tradeoff is the difference between reading a TR4 or a Bethesda III result as a diagnosis and reading it as a position on a risk curve.

This article is educational and is not medical advice; decisions about a specific nodule belong to a person and their own clinician. The practical takeaway is interpretive: a TI-RADS level answers whether to biopsy, a Bethesda category answers what the cells suggest, and neither number is the final word that histopathology or longitudinal follow-up ultimately provides.