One structure, one number, five minutes. The median nerve found among the flexor tendons by tilting the transducer, its cross-sectional area taken on the frozen frame, and that number read against a published cut-off.
Sobre a Dra. Mariz
I am Dr. Fernando Mariz, a gynecology and pelvic surgery physician practicing in New York City. Before my medical career, I served in the U.S. Marine Corps, where I developed the discipline, focus, and steadiness that continue to shape the way I care for patients today. At Maiden Lane Medical, my work covers women’s health, preventive care, sonography, pelvic pain, abnormal uterine bleeding, and minimally invasive gynecologic procedures.
Sobre o Dr. Avila
I am Dr. Jailyn Avila, an emergency medicine physician, educator, and digital health innovator based in Southern California. My clinical work is rooted in emergency care, while my teaching focuses on point-of-care ultrasound, residency education, and practical training for physicians. Through my roles with Emergent Medical Associates, UHS SoCal MEC Emergency Medicine Residency, Core Ultrasound, and FemInEM, I work at the intersection of patient care, medical education, and accessible digital learning.
Resposta rápida
A carpal tunnel ultrasound is one measurement: the cross-sectional area of the median nerve, taken in transverse at the wrist, where the nerve is at its widest. You find the nerve by tilting the transducer until the flexor tendons darken and drop out of the image and the nerve does not, freeze on the largest cross-section, and take the area with the ellipse or the trace tool. Published cut-offs cluster between 8.5 and 10 mm² at the carpal tunnel inlet, and the AANEM rates the measurement as an accurate diagnostic test for carpal tunnel syndrome at its highest evidence level. The scan on this page runs the whole sequence in five minutes on a general-purpose transducer, with the operator saying on camera exactly what a higher-frequency probe would have bought him. My advice on a first watch: take 1:40 to 3:10 end to end — that is the identification and the measurement, and everything else on the page hangs off those ninety seconds.
Looking for hardware rather than the exam? The musculoskeletal hub compares the models used for nerve and small-parts work, and the Página do produto D3Ultra carries the full specification and current price.
The decision chain
Five moves from a wrist to a number
This scan is not a tour of anatomy. It is one structure and a short chain of decisions, each of which has to hold for the next one to mean anything. Click any step to jump the recording to it.
01SeparateTilt until the tendons disappearIn cross-section a nerve and a tendon look alike. Tilting the transducer settles it: tendons are far more sensitive to anisotropy and darken off-axis, so the structure still visible at an angle is the nerve.anisotropytilt on and off axisB-mode1:4002MaximiseSweep for the widest pointThe area you measure is only meaningful if it is the largest one. A short proximal-to-distal sweep finds it, and the frame is frozen there rather than wherever the probe happened to stop.proximal to distalfreeze on the maximum2:2503MeasureEllipse, or trace it by handTwo of the six tools in the measurement menu return an area. The ellipse is faster; the trace follows a nerve that is not elliptical. Both are taken inside the hyperechoic rim, not around it.AREA/CIRCUMTRACE2:5004DecideRead the area against a cut-offOne number, one comparison. The frozen frame here returns 0.08 cm² — 8 mm², under every cut-off in common use, which is what a normal median nerve looks like.cross-sectional arearule-out3:1005ConfirmWiggle, then compare proximallyTwo cheap manoeuvres when the number sits close to the line: fingers moving should carry the nerve with them, and a nerve that widens as it runs distally is doing the opposite of what a healthy one does.finger wiggleproximal comparison4:06
The confidence ladder, in his numbers
The ladder is the escalation the presenting clinician sets out at 3:32 — his point being that a value sitting just over a cut-off is a weaker call than one well past it. The threshold this page states in its own voice is the published one, and it is cited in Sources.
Timestamped
Key moments, with the settings
The third column is the part that usually goes missing from a scan video. Two things are worth noticing in it. The first is that the preset, the depth and the focus positions are set once, at 1:04, and never touched again — unlike a multi-window study, a nerve scan is one setup. The second is what the interface does não show: this recording has the parameter overlay switched off, so gain, dynamic range and transmit frequency are never displayed. The frequency ceiling is stated out loud instead, at 0:12.
| Tempo | What is on screen | Transducer, preset, on-screen state |
|---|---|---|
| 0:00 | The probe you have, and the one you would rather haveCarpal tunnel syndrome is framed as a compression neuropathy of the median nerve at the wrist. The presenting clinician states on camera that this unit tops out at 10 MHz and that 12, 15 or 18 MHz would resolve the nerve better — then makes the argument the rest of the recording has to earn: a general-purpose transducer is enough to make this call. | SX-6CT · MSK preset · D 20 mm · LIVE · software V 3.6.74 |
| 0:36 | Positioning the wristGel is already on the transducer. The forearm laid flat on a table is the preferred setup; holding the arm works too, and is what he says he does most of the time, because the scan is short enough not to need the furniture. | SX-6CT · MSK preset · D 20 mm · LIVE |
| 1:04 | Preset, focus, gain and depthThe general musculoskeletal preset is selected — a dedicated nerve preset if the unit has one, MSK if not. Focus, gain and depth are named as the three things to set before you start looking, and on this recording they are set once and never touched again. | MSK preset · D 20 mm · two focus markers at ≈4 mm and ≈9 mm · rail: Gain+, Gain−, Depth, Focus, Dyn., Harmonic, Denoise, B Mode |
| 1:40 | Anisotropy: telling nerve from tendonThe single most useful beat in the recording. The transducer is tilted back and forth across the flexor tendons and structures wink in and out of the image. Tendons are the sensitive ones; the nerve is not. The structure that stays visible through the tilt is the median nerve. | SX-6CT · MSK preset · D 20 mm · LIVE · B-mode, no Doppler |
| 2:25 | Finding the largest cross-sectionA short sweep proximal and distal looking for the point where the nerve is widest, then freeze. The structure that did not wink out on the tilt is identified on the frozen frame as the median nerve. | MSK preset · D 20 mm · LIVE → FREEZE · cine buffer 100/100 |
| 2:50 | Ellipse or traceThe measurement menu is opened over the frozen frame: LENGTH, AREA/CIRCUM, TRACE, Depth, ANGLE, Area Ratio. Two of those six return a cross-sectional area — an ellipse fitted to the nerve, or an outline drawn around it by hand. | FREEZE · Meas ▸ AREA/CIRCUM or TRACE |
| 3:10 | Reading the numberThe closed trace returns AREA 0.08 cm² on screen. The presenting clinician gives his own cut-off in the same breath: 0.09 cm², which is 9 mm², at close to full sensitivity and a specificity he puts in the high 80s to low 90s. | FREEZE · trace closed · on-screen AREA 0.08 cm² |
| 3:32 | When the number is well over the lineThe escalation, and the part most threshold tables leave out: a value a little over the cut-off is a weaker call than 11, 12 or 13 mm². He names 12 mm² as the point where he stops hedging. | FREEZE · measurement retained on screen |
| 4:06 | Two confirmatory manoeuvresAsk the patient to wiggle their fingers and the nerve should shift with the tendons; one that stays put is a second sign. Then scan up to about 10 cm proximal and compare areas — a nerve should narrow as it runs distally, so an increase, which he puts at 40% or a factor of 1.4, points the other way. | SX-6CT · MSK preset · D 20 mm · LIVE |
| 4:49 | The close, and a caption that disagrees with itThe exam closes on the cut-off used as a rule-out. A burned-in caption then puts a second figure on screen — that 10 mm² could be the better balance of sensitivity and specificity — next to two reference images captured on a different machine. | Reference images and captions · no live scanning |
Transcribed from the device interface visible in the recording. Depth is read off the on-image scale, which is marked at 0, 5, 10 and 15 mm with the field ending at 20 mm; the two focus carets sit at roughly 4 mm and 9 mm. Software build V 3.6.74.
Reading the images
What this scan shows
In a transverse view of the anterior wrist the median nerve is a speckled oval sitting superficially among the flexor tendons, and at first pass it looks like one of them. The separation is not a matter of shape but of behaviour. Tendons are strongly anisotropic: angle the transducer a few degrees off perpendicular and their fibrillar echoes reflect away from the probe and the tendon goes black. Nerves are far less sensitive to that. So the manoeuvre at 1:40 is the whole identification — rock the transducer back and forth and watch which structure refuses to disappear. That one is the nerve.
Where you measure matters as much as what you measure. The area is taken at the carpal tunnel inlet, at the level of the pisiform, in a true transverse plane, and at the point along the tunnel where the nerve is widest — which is why the recording sweeps a short distance proximal and distal before freezing rather than measuring the first clean image it finds. On the frozen frame two of the six measurement tools return an area: an ellipse fitted to the nerve, or a trace drawn around it by hand. Every Class I study behind the AANEM guideline used direct tracing, and the outline goes inside the bright rim of the nerve, not around it. Trace the rim in and the number comes back inflated.
Then there is the number itself, and this recording is a good example of why a page like this should not simply repeat one. Three different cut-offs appear on the same asset: 9 mm² spoken at 3:10, 10 mm² in the video description, and a burned-in caption near the end offering 10 mm² as the better balance of sensitivity and specificity. The literature explains the spread rather than settling it. A systematic review of 41 studies and 2,504 nerves put the mean inlet area in healthy people at 8.74 mm², so any diagnostic threshold has to sit just above the top of the normal range — and the Class I studies behind the AANEM guideline used 8.5 to 10 mm², while a meta-analysis of 3,995 wrists found 9 mm² the best single criterion at 87.3% sensitivity and 83.3% specificity. Read that spread the way it is meant to be read: a threshold is a dial between catching everything and calling too much, not a line in nature.
Which is exactly why the two manoeuvres at 4:06 are worth the extra ten seconds when a number lands close to the line. Asking the patient to move their fingers should carry the nerve along with the tendons; a nerve tethered in a tight tunnel does not travel. And comparing the wrist against the mid-forearm turns one absolute number into a ratio, which is more forgiving of body size and of a hurried caliper. That ratio has a name and a figure in the literature — roughly 1.0 in asymptomatic volunteers, an average of 2.1 in patients, and a value of 1.4 catching every affected patient in the series that defined it.
The last variable is the hardware, and the operator is unusually honest about it in the first thirty seconds. Nerve work is a resolution problem: fascicles are sub-millimetre, and 12, 15 or 18 MHz resolves them where 10 MHz gives you an oval. What 10 MHz does give you is enough to find the nerve, freeze on its widest point and take an area — which is the measurement the diagnosis actually turns on. That is a real distinction worth keeping straight when choosing a probe: the cheaper frequency costs you the texture inside the nerve, not the number around it.
Verbatim
Full transcript
Transcript — 5:27, 11 passages
Transcribed from the recording and edited for readability; square brackets mark an editorial clarification, and every timestamp jumps the video.
Setup
0:00[Introducing himself and the exam.] Today I’d like to show you how to evaluate for carpal tunnel syndrome — a compression peripheral neuropathy of the median nerve at the wrist, or at the carpal tunnel. So ideally you’re using as high a frequency probe as you have access to, a linear probe. But today I’m going to actually use a Suresult D3Ultra, which is a dual-purpose probe and it only goes up to 10 megahertz. Now, ideally 12 megahertz is better, 15 megahertz is better, 18 megahertz is better — but use what you have. And I want to show that you can use a general-purpose probe to diagnose carpal tunnel syndrome.
0:36So I’ve already applied gel to the transducer, and I have my patient in my preferred position. I find this position — if you can pull a table close by, or if you have a table in your office where the patient can just lay their arm down, that’s best. Of course, this is such a quick and simple scan you could just hold their arm and scan, which is what I typically do. It’s so quick and easy.
1:04Okay, so I have things set up and we can go. I like to use sort of the presets — general musculoskeletal preset. Sometimes your unit — I’m not sure what unit you’re using — may have a nerve preset, but even a general musculoskeletal preset should work. So, as you can see from what we’re seeing, there’s a heck of a lot going on, right? And of course you want to make sure your focus points are set correctly, the gain is set correctly and depth is set correctly, so you can get as best a picture as you can.
Finding the nerve
1:40So I’m going back and forth here, and you’ll see that there’s so many structures. What the heck are we looking for? Well, as I tilt back and forth, things wink in and out of existence — this is an imaging artifact called anisotropy, and tendons are more sensitive than nerves. Now in cross-section a nerve and a tendon look very similar, but since nerves are less sensitive to anisotropy, the structure which is more persistent is going to be the nerve. So as I go back and forth, you’ll see that there’s one thing that looks like a tendon, just like the other tendons, but it doesn’t go black as I tilt back and forth. And that’s going to be the nerve.
2:25So I can go back and forth a little bit — a little bit proximal, a little bit distal — and I look for where that nerve is at its largest. Then I can pause, and I can get out my measurement tools. So if you were looking at the images, the structure that wasn’t winking in and out of existence is this structure here. So that’s going to be our median nerve.
Measuring
2:50So I can take our measurement tools, and I can either use an ellipse — an ellipse will typically work — or I can use a trace, and I can trace the shape of the nerve. And then I look at the cross-sectional area.
3:10If you use a threshold of 0.09 centimetres squared, or 9 millimetres squared, that gives you almost 100% sensitivity — you know, high 90s, 100% sensitivity — and around high 80s, low 90s specificity. So that will catch — well, actually, sorry, 100% sensitivity — that will basically catch all carpal tunnel syndrome. And it was just that quick and easy.
3:32Now, as I mentioned, the specificity isn’t quite 100%; it’s sort of hovering around 90% in the studies. If you want to improve your confidence — you know, the larger it is, the more confident you can be in your diagnosis. So if, instead of it just being a little over that 9 millimetres squared threshold, if it’s 11 millimetres squared, 12 millimetres squared, 13 millimetres squared, then you can be more confident in your diagnosis, particularly if it’s over 12 millimetres squared.
Confirming
4:06Another finding is if you have the patient wiggle their fingers, you should see the nerve sort of wiggle back and forth. Of course, you should be practising on normals — and if it doesn’t wiggle, that can be another sign, to improve your confidence in your diagnosis. The other thing you can do is you can scan the nerve more proximal, maybe up until about 10 centimetres more proximal, and you can compare that cross-sectional area. Nerves should only get smaller as you go more distal, but if the nerve has increased in size, that increases the likelihood of carpal tunnel syndrome — in particular, if the increase in size is larger than 40%, so a factor of 1.4.
Fechar
4:49But even if you don’t do those extra things, just using that cutoff of 9 millimetres squared, again, has excellent power to rule out that condition, so you can start looking for other conditions. And did you see how quick and easy that was? So get practising, get using your ultrasound, and best of luck helping your patients. Cheers.
5:13[Closing cards, 5:13–5:27 — no speech.]
Hardware
Device and settings
The left column is the published specification. The right column is what the interface actually showed while this scan was being run, and that is the column I would work from if you are reproducing it. The short version: linear side, musculoskeletal preset, 20 mm of depth, two focus carets in the top half of the image, and nothing touched after 1:04. If this exam is one among many, a multi-geometry head covers it without an argument. If nerve and small-parts imaging is almost all of your week, a multi-geometry head is not a good fit: buy a dedicated high-frequency linear probe instead, because the frequency ceiling is the thing you will hit every day.
From the published specification
Suresult D3Ultra
Three geometries in one head; the linear one is what this scan uses — $2,976
- Linear side7.5 and 10 MHz, the frequencies that matter here
- Linear depth20, 40, 60 or 100 mm · 40 mm footprint
- Grey scale256 levels · gain adjustable 30 to 105 dB
- Aperture192 elements driven on 64 channels
- Also in the headConvex and phased geometries, 3.2 and 5.0 MHz
- Imaging modesB and M, plus colour, power and pulsed-wave Doppler
- Tamanho263 g · 156 by 65 by 20 mm
- EnduranceAround two hours of continuous scanning
- Connects toiOS, Android or Windows over dual-band Wi-Fi
What the interface showed while this scan ran
- HeaderSX-6CT GRCEKR009 · MSK preset
- GeometryLinear side — the image is rectangular for the whole recording
- Profundidade20 mm · scale marked 0 / 5 / 10 / 15 mm
- FocoTwo carets, at roughly 4 mm and 9 mm
- FrequênciaNot displayed — a 10 MHz ceiling is stated on camera at 0:12
- ModeB-mode throughout · no M-mode, no Doppler
- MediçãoMeas ▸ AREA/CIRCUM or TRACE · frozen frame reads AREA 0.08 cm²
- RailGain+, Gain−, Depth, Focus, Dyn., Harmonic, Denoise, B Mode
- SoftwareV 3.6.74 · cine buffer 100/100 on freeze
- Operator notePreset, depth and focus are set at 1:04 and unchanged to the end
Same device, other structures
Related scans
4:34De Quervain’s tenosynovitis at the radial styloidThe other wrist scan worth knowing, and one of the conditions that sends a patient to a carpal tunnel clinic in the first place: the first extensor compartment, in transverse.Watch in the Library →
3:33Plantar fascia: measuring at the insertionThe same measurement discipline applied to a different structure — find the thickest point, freeze, and read one number against a published threshold.Watch in the Library →
14:09eFAST trauma exam on the same handheld probeThe long-form study on the same hardware: five windows, the presets and depths changing five times, and the full transcript.Read the walkthrough →All the clinical clips, filterable by anatomy and probe, are in the Scan Library.
Asked on this search
Carpal tunnel questions
Can carpal tunnel syndrome be diagnosed with ultrasound?
Yes. The American Association of Neuromuscular and Electrodiagnostic Medicine rates measurement of median nerve cross-sectional area at the wrist as an accurate diagnostic test for carpal tunnel syndrome, at its highest evidence level. Across the four Class I studies behind that guideline, sensitivity ran from 65% to 97% and specificity from 73% to 98% — a spread wide enough that the scan sits alongside the clinical picture and nerve conduction studies rather than replacing them. What it adds that electrodiagnosis cannot is the anatomy: a ganglion, a tenosynovitis or a bifid nerve is visible in the same ten seconds.
Source: AANEM evidence-based guideline — neuromuscular ultrasound for carpal tunnel syndrome. Accessed September 5, 2026.
How do you identify the median nerve on ultrasound?
By tilting. In a transverse view of the anterior wrist the median nerve sits among the flexor tendons and looks much like them — a speckled, honeycomb oval. Tendons are far more sensitive to anisotropy than nerves, so angling the transducer off-perpendicular makes the tendons darken and drop out of the image while the nerve stays bright. The structure still there at an angle is the nerve. That single manoeuvre, at 1:40 in the walkthrough above, is the whole identification step.
Stated by the presenting clinician in the recording above.
What is a normal median nerve cross-sectional area?
A systematic review pooling 41 studies and 2,504 nerves put the mean cross-sectional area at the carpal tunnel inlet at 8.74 mm² in healthy people, with a 95% confidence interval of 8.45 to 9.03 mm². Diagnostic cut-offs sit right on top of that range: the Class I studies behind the AANEM guideline used 8.5 to 10 mm², and a meta-analysis of 28 studies covering 3,995 wrists found 9 mm² or more the best single criterion, at 87.3% sensitivity and 83.3% specificity. The frozen frame in the walkthrough reads 0.08 cm² — 8 mm² — which is under all of them.
Source: Normative reference values of the median nerve cross-sectional area in healthy individuals. Accessed September 5, 2026.
Where in the wrist is the median nerve measured?
At the carpal tunnel inlet, the level of the pisiform, in a true transverse plane; the outlet at the hook of the hamate is the second measured level. The area is taken where the nerve is widest, which is why the walkthrough sweeps a short distance proximal and distal before freezing rather than measuring the first clean image. Every Class I study behind the AANEM guideline used direct tracing, drawn inside the hyperechoic rim of the nerve — trace around the rim instead and the number comes back too big.
Source: AANEM evidence-based guideline — neuromuscular ultrasound for carpal tunnel syndrome. Accessed September 5, 2026.
Does comparing the wrist with the forearm add anything?
It does, and it is the manoeuvre at 4:06 in the walkthrough. Dividing the nerve’s area at the wrist by its area in the mid-forearm gives the wrist-to-forearm ratio, which is close to 1.0 in asymptomatic volunteers and averaged 2.1 in patients presenting with carpal tunnel syndrome. In that series a ratio of 1.4 detected every affected patient. It is worth the extra ten seconds whenever a single wrist measurement lands close to the cut-off.
Source: The ultrasonographic wrist-to-forearm median nerve area ratio in carpal tunnel syndrome. Accessed September 5, 2026.
Can a carpal tunnel ultrasound be done on a handheld probe?
It is done on one in the recording above, and the operator says plainly what he is giving up: the unit tops out at 10 MHz, and 12, 15 or 18 MHz would resolve the fascicles better. Resolution is the whole argument on a nerve scan, so buy the highest linear frequency you will actually use. My own line is this — a general-purpose head is enough to take a cross-sectional area and make the call, and if nerve and small-parts work is most of your week, a dedicated high-frequency linear probe is the better purchase.
Stated by the presenting clinician in the recording above.
Provenance
Fontes
- AANEM evidence-based guideline — neuromuscular ultrasound for carpal tunnel syndromeCartwright, M. S. et al. Muscle & Nerve, 2012;46(2):287–293; reaffirmed September 2017. Rates median nerve cross-sectional area at the wrist as an accurate diagnostic test (Level A); the Class I cut-offs range from 8.5 to 10 mm² and all used direct tracing. Accessed September 5, 2026.
- Ultrasonography for diagnosing carpal tunnel syndrome: a meta-analysis of diagnostic test accuracyTai, T. W. et al. Ultrasound in Medicine & Biology, 2012;38(7). Twenty-eight trials, 3,995 wrists. Source of the 9 mm² inlet criterion quoted on this page, at 87.3% sensitivity and 83.3% specificity. Accessed September 5, 2026.
- Normative reference values of the median nerve cross-sectional area in healthy individualsNg, A. J. T. et al. Scientific Reports, 2022;12:9217. Systematic review of 41 studies and 2,504 nerves. Source of the 8.74 mm² pooled mean at the carpal tunnel inlet, CI 8.45–9.03, and of the inlet and outlet landmarks. Accessed September 5, 2026.
- The ultrasonographic wrist-to-forearm median nerve area ratio in carpal tunnel syndromeHobson-Webb, L. D. et al. Clinical Neurophysiology, 2008;119(6). Source of the wrist-to-forearm ratio figures: 1.0 ± 0.1 in asymptomatic volunteers, 2.1 ± 0.5 in patients, and 100% sensitivity at a ratio of 1.4. Accessed September 5, 2026.
- Meta-analysis on the performance of sonography for the diagnosis of carpal tunnel syndromeDescatha, A. et al. Seminars in Arthritis and Rheumatism, 2012;41(6). Pools sensitivity and specificity threshold band by threshold band, which is where the trade-off across the 8.5–13 mm² range comes from. Accessed September 5, 2026.
- Página do produto Suresult D3UltraLinear frequency options and their depth settings, aperture and channel counts, grey levels, gain range and weight, taken from the published product attribute table. Accessed September 5, 2026.
- Suresult channel — the recording on this page“Carpal Tunnel Syndrome Evaluation with Handheld Ultrasound”, published 13 October 2025. Duration 5:27. The transcript, the key moments and every scan frame on this page are drawn from it. Accessed September 5, 2026.
How much frequency do you actually need for nerve work?
Ten megahertz is enough to find the median nerve, freeze on its widest point and take the area the diagnosis turns on. Twelve to eighteen shows you the fascicles inside it. Tell us which scans fill your week and you will get a direct answer on where the extra frequency earns its money and where it does not.
