How to Dictate a Carotid Duplex: Velocity Criteria, the Ratio, and the String Sign

The carotid duplex is the most formulaic report in ultrasound, and residents mistake that for a weakness. It is the opposite: the surgical trials that justify endarterectomy were built on specific stenosis thresholds, so a carotid report is only useful if it maps velocities to those same categories, the same way, every time. Your job is not creativity. Your job is to apply the table honestly and to know the three situations where the table lies.

The consensus criteria

The widely used framework is the 2003 Society of Radiologists in Ultrasound consensus criteria, built around three numbers: ICA peak systolic velocity (PSV), ICA end-diastolic velocity (EDV), and the ICA/CCA PSV ratio.

Memorize the two PSV cutpoints — 125 and 230 cm/s — and the ratio cutpoints 2.0 and 4.0. Everything else in the report is built on them. If your lab has validated its own thresholds, use the lab's; the principle is consistency, not the specific table.

Report ranges, not percentages

The categories exist because duplex cannot distinguish 62% from 68%. Dictate "50–69% stenosis" or "70% or greater stenosis, not near-occlusive" — the ranges the trials used — and never a single invented percentage. "Approximately 65% stenosis" sounds more precise and is strictly less honest. The vascular surgeon's decision boundaries are the category edges; give them the category.

When the ratio outranks the velocity

The raw PSV assumes a normal pump and a normal contralateral side. Three situations break it:

1. Contralateral high-grade stenosis or occlusion inflates velocities on the open side (compensatory flow). A PSV of 140 cm/s opposite an occlusion may represent a <50% lesion; the ratio corrects for it because the CCA carries the same elevated flow.
2. Low cardiac output, severe aortic stenosis, or tandem lesions depress all velocities — a 70%+ stenosis can hide under a PSV of 180. Again, the ratio survives because numerator and denominator fall together.
3. Long or tortuous segments make angle correction unreliable; say so rather than reporting a number you don't believe.

When PSV and ratio disagree, say both, say which category you favor and why, and describe the plaque. A discrepant exam dictated as a clean category is how patients get operations they don't need.

Read the waveforms, not just the peaks

Before any velocity means anything, confirm you are in the vessel you think you are in. The two branches of the CCA have opposite personalities: the ICA is a low-resistance vessel — broad systolic peak, continuous forward diastolic flow, because the brain wants blood all cardiac cycle — while the ECA is high-resistance, with a sharp peak and little diastolic flow, and it has branches in the neck. When the bifurcation anatomy is ambiguous, the temporal tap settles it: tapping the superficial temporal artery puts visible oscillations on the ECA tracing. Measuring a stenosis in the ECA and reporting it as the ICA is a classic, fully preventable error, and the waveform plus the tap prevents it.

The waveforms also diagnose disease you cannot see:

• A parvus tardus pattern — slow systolic upstroke, rounded peak — in the CCA or ICA points to an obstruction proximal to your sample: innominate or proximal CCA stenosis hiding under the clavicle.
• A high-resistance ICA waveform with little or no diastolic flow points to obstruction distal to your sample — distal ICA or intracranial occlusion — and a technically "normal" PSV with that waveform is not a normal exam.

One sentence on waveform character costs nothing and is frequently the only clue the extracranial exam gives you about disease outside its field of view.

Near occlusion and the string sign

Near occlusion is the category everyone gets wrong, because the velocities fall as the lumen approaches closure — the jet through a hairline channel can be paradoxically low. The grayscale and color findings carry the diagnosis: a small, threadlike distal ICA with a trickle of flow — the string sign — often needing low PRF and power Doppler to detect at all.

The distinction between near occlusion and occlusion is the whole report. An occluded ICA is managed medically; a near-occluded ICA may still be a revascularization candidate. If duplex cannot confidently separate them, say exactly that and recommend CTA or MRA. "No flow identified; near occlusion not excluded — recommend CTA" is a better sentence than a confident call the angiogram will reverse.

Vertebrals: direction is the finding

Every carotid report says something about both vertebral arteries, and the minimum is flow direction:

• Antegrade — normal.
• Retrograde — subclavian steal: the arm is perfusing itself through the brain's circulation, and the report should say "retrograde right vertebral flow, consistent with subclavian steal; correlate with arm blood pressures."
• To-and-fro (partial steal) — early systolic deceleration or alternating flow is latent steal; it may become complete with arm exercise.

A steal is usually asymptomatic and usually matters only with symptoms or a planned LIMA graft — but the direction takes one second to dictate and omitting it is the most common deficiency in resident carotid reports.

Plaque deserves a sentence, not an essay

Describe the worst plaque briefly: location, calcified versus hypoechoic, smooth versus irregular. Heavily calcified plaque with shadowing can hide the velocity jet entirely — if the sample volume couldn't interrogate the stenosis, the report must say the estimate is limited. And keep one habit from day one: the duplex field of view includes the thyroid, and a carotid study is a common source of incidental thyroid nodules. Don't adopt them into the carotid impression unworked — size them and route the follow-up through ACR TI-RADS rather than reflexively recommending a dedicated ultrasound for every 4 mm colloid nodule.

A model report

EXAMINATION: Carotid duplex ultrasound, bilateral
HISTORY: 71-year-old male, left hemispheric TIA.
TECHNIQUE: Grayscale, color, and spectral Doppler of the
cervical carotid and vertebral arteries.
COMPARISON: None.
FINDINGS:
Right: Scattered calcified plaque, proximal ICA. PSV
98 cm/s, EDV 28 cm/s. ICA/CCA ratio 1.4. Antegrade
vertebral flow.
Left: Irregular, predominantly hypoechoic plaque at the
carotid bifurcation extending into the proximal ICA.
PSV 312 cm/s, EDV 118 cm/s. ICA/CCA ratio 4.8.
Antegrade vertebral flow.
IMPRESSION:
1. Left proximal ICA stenosis, 70% or greater but not
   near-occlusive, by SRU consensus criteria
   (PSV 312 cm/s, ratio 4.8).
2. Right ICA plaque with less than 50% stenosis.
3. Antegrade vertebral flow bilaterally.

The impression names the criteria and shows the numbers that drove the category — the vascular surgeon can audit the call in five seconds. Keep the normal carotid ultrasound template as your scaffold, and the rest of the guideline calculators for what the incidentals turn up. Or run a carotid case through the live demo — dictate two velocities and a ratio, and watch the consensus category, correctly phrased as a range, land in the impression on its own.