Ultrasound Calipers Disagree After Export: Check Region Calibration

A distance measured on the ultrasound console reads 4.6 cm, yet when a radiologist or technologist measures the same frozen structure in PACS after export, the workstation reports a different number, displays raw pixels with no physical units, or shifts values following a viewer upgrade. The same question routinely surfaces during post-repair acceptance when a biomedical engineer or independent service organization (ISO) returns a scanner after a board or transducer repair and asks whether the exported measurements still agree with the cart. Treating that discrepancy as an immediate hardware failure wastes replacement parts and bench time; treating it as trivial operator variation risks overlooking an export-format strip, an on-scanner geometric fault, or a documented PACS viewer regression. ^{1 3 8}

This verification guide separates on-image region-calibration scale from adjacent DICOM decisions in our series. If discrete numeric measurements or OB/vascular tables are missing from PACS or the reporting package altogether, start with our ultrasound measurements missing in PACS SR path guide—that decision covers whether coded measurement values arrived at all (burned-in annotations versus Structured Reporting), whereas this guide covers why calipers drawn on an exported ultrasound image produce a different scale or unit. If a saved multi-frame loop plays back at the wrong tempo, use our ultrasound cine frame timing evidence guide. Separate decisions not covered here include whether earlier lossy compression history is disclosed in the header, whether a 3D/4D study exports as a true volume SOP class versus a rendered cine clip, and whether resend workflows preserve or replace SOP Instance UID identity.

What does the US Region Calibration Module actually store?

In the DICOM Standard (PS3.3 Section C.8.5.5, reviewed in the current 2026d online edition and confirmed item-by-item against the 2022a edition), the US Region Calibration Module exists because a single ultrasound image frame frequently contains multiple independent display areas—such as dual B-mode views, quad-screen loops, B-mode paired with M-mode, or a 2D tissue sector above a pulsed-wave (PW) spectral Doppler trace and ECG waveform. ¹ Each area on the canvas can represent a different physical parameter—spatial distance, blood velocity, time, or volume—and carries its own pixel coordinate geometry. ¹

There is a two-level conformance distinction that every PACS administrator and clinical engineer should verify before calling an exported file defective. In the standard IOD module tables for traditional ultrasound—PS3.3 Section A.6.4 (Table A.6-1, Ultrasound Image IOD) and Section A.7.4 (Table A.7-1, Ultrasound Multi-frame Image IOD)—the US Region Calibration Module (C.8.5.5) has Usage = U (User Optional). ² A scanner is not required by the base IOD table to include the module in every exported ultrasound image. However, when the US Region Calibration Module is present in the SOP Instance, Table C.8-17 makes Sequence of Ultrasound Regions (0018,6011) a Type 1 sequence requiring one or more items, and makes the core bounding, unit, and delta attributes inside each sequence item Type 1 (mandatory and non-empty). ^{1 2}

Why do whole-image pixel-spacing assumptions fail on ultrasound?

Engineers accustomed to CT, MR, or digital radiography often expect every DICOM image to carry a single top-level Pixel Spacing (0028,0030) attribute expressed in millimeters per row and column. Traditional DICOM ultrasound does not work that way. As DICOM committee contributor David Clunie summarized in a February 2024 3D Slicer support discussion, in traditional DICOM ultrasound "the pixel size information is specific to regions" and, "if present at all, may be in the US Region Calibration Module, specifically in PhysicalDeltaX and PhysicalDeltaY, with corresponding units." ⁶ When we inspected the full text of PS3.3 Section C.8.5.5 in both the 2022a and current 2026d editions, the standard page did not mention Pixel Spacing (0028,0030) once—nor does it define any whole-image fallback when (0018,6011) is absent. ¹ Some ultrasound datasets do also carry a top-level spacing value—one ITK forum participant notes that "some data sets may use Pixel Spacing"—but the ultrasound IOD module tables do not define one, so where it appears it is not a substitute for checking the region under the caliper. ^{2 7}

General-purpose imaging toolkits do not treat ultrasound region deltas as whole-image voxel spacing either, and there are sound reasons for that. In an Insight Toolkit (ITK) forum thread asking whether Physical Delta X (0018,602C) should be used as image spacing, a SimpleITK developer replied that it is not necessarily geometric spacing (it can be time for Doppler), and another participant added that ultrasound files often contain several, sometimes overlapping, regions whose deltas can be zero or negative—"it is for measurements in particular region, not whole image." Later posters reported that SimpleITK's spacing ignored the value. ⁷ The standard itself requires a negative Physical Delta Y in spectral Doppler regions. ¹ Consequently, a viewer or research pipeline that relies on generic whole-image spacing rather than parsing Sequence of Ultrasound Regions (0018,6011) at the cursor coordinates will either fall back to raw pixel counts or apply the wrong region's scale. ^{4 5 6 7}

First check: the same stored object, measured in both places

Before anyone suspects a faulty beamformer board, replaces a transducer, or opens a PACS database ticket, run a strict same-object, two-place read-only verification. Do not compare a live re-scan on the patient against a historical PACS frame, and do not compare a burned-in measurement summary or a Structured Report value against a newly drawn freehand caliper without first confirming you are inspecting the exact same stored frame. ^{3 5}

1. Lock the exact SOP Instance and frame number: Record the SOP Instance UID (0008,0018), SOP Class UID (0008,0016), and—if the object is a multi-frame loop—the exact frame index where the discrepancy was reported. Preserve both the local study on the ultrasound cart and the archived PACS instance; follow our DICOM storage commitment deletion decision guide so no source study is purged while troubleshooting is open.
2. Compare burned-in console caliper text (if present) against a fresh console caliper on the same frozen frame: On the ultrasound console's local review screen, open the stored frame and place a linear distance caliper between two unambiguous anatomical or phantom landmarks (or re-measure the exact endpoints of an existing burned-in caliper marker). Record the console value and displayed units (for example, 2.50 cm or 25.0 mm). ^{3 5}
3. Measure the identical pixel endpoints in the receiving PACS viewer: Open the same SOP Instance UID and frame in the PACS or review workstation, place the viewer's distance tool on the exact same landmark pixels, and record the displayed number, unit label (cm, mm, px, or blank), and whether the viewer highlights an active calibration region boundary. ^{4 5}
4. Dump the header read-only to inspect (0018,6011): Using the PACS viewer's DICOM header inspector or a read-only command-line dump of a workstation copy of the object the receiver actually holds, verify whether Sequence of Ultrasound Regions (0018,6011) is present, how many sequence items exist, and what bounds, units, and deltas each item declares. ^{1 5}
5. Check viewer version and active recall notices: Record the exact PACS viewer product name and software build (for example, checking whether the workstation runs Merge PACS or Merge OrthoPACS V9.5 covered by FDA Class 2 recall Z-3004-2026) and test whether a second independent region-aware viewer reading the same file agrees with the console or with PACS. ^{3 5}

Reading the region evidence: bounds, deltas and units

To see how the header attributes govern the caliper math, inspect the sequence items inside (0018,6011). Every sequence item defines a bounding rectangle using zero-based pixel coordinates: Region Location Min x0 (0018,6018), Region Location Min y0 (0018,601A), Region Location Max x1 (0018,601C), and Region Location Max y1 (0018,601E), where 0\0 is the top-left pixel of the image matrix. ¹ Inside that box, Physical Units X Direction (0018,6024) and Physical Units Y Direction (0018,6026) state the units via 16-bit hexadecimal enumerated values, and Physical Delta X (0018,602C) and Physical Delta Y (0018,602E) state the physical increment per pixel step. ¹

Synthetic Header & Caliper Illustration (No Patient Data): In a synthetic 960 × 720 duplex ultrasound frame constructed from PS3.3 Section C.8.5.5, Philips Lumify R5.0 conformance tables (HSDP-1299212), and the Medical Connections DicomObjects geometry example, Item 1 (upper 2D tissue sector) sets Region Spatial Format (0018,6012) = 0001H (2D), Region Data Type (0018,6014) = 0001H (Tissue), Region Flags (0018,6016) = 3, pixel bounds x0=80, y0=40, x1=880, y1=440, Physical Units X/Y (0018,6024)/(0018,6026) = 0003H (cm), and Physical Delta X/Y (0018,602C)/(0018,602E) = 0.05. Item 2 (lower PW spectral Doppler strip) sets Region Spatial Format = 0003H (Spectral), Region Data Type = 0003H (PW Spectral Doppler), pixel bounds x0=80, y0=460, x1=880, y1=680, Reference Pixel x0/y0 = 0 / 110, Physical Units X = 0004H (seconds), Physical Units Y = 0007H (cm/sec), Physical Delta X = 0.004, and Physical Delta Y = -0.80. ^{1 4 9}— Synthetic two-region DICOM header illustration modeled on DICOM PS3.3 C.8.5.5, Philips Lumify R5.0 HSDP-1299212, and Medical Connections DicomObjects KB

Walk through the exact arithmetic on Item 1 above using the worked geometry documented by Medical Connections: ⁴ suppose a user places a caliper inside the upper 2D region from pixel (200, 150) to pixel (230, 190). Both endpoints lie inside the Item 1 bounds [80..880, 40..440]. The pixel displacement is Δx = 30 pixels horizontally and Δy = 40 pixels vertically, giving an uncalibrated Euclidean length of √(30² + 40²) = 50 pixels. ⁴ Because Item 1 declares Physical Units X/Y = 0003H (cm) and isotropic Physical Delta X = Physical Delta Y = 0.05 cm/pixel (which equals 0.5 mm/pixel), the physical displacement is:

• Calibrated horizontal component: 30 pixels × 0.05 cm/pixel = 1.50 cm (15.0 mm). ^{1 4}
• Calibrated vertical component: 40 pixels × 0.05 cm/pixel = 2.00 cm (20.0 mm). ^{1 4}
• Calibrated Euclidean distance: √(1.50² + 2.00²) = 2.50 cm (25.0 mm). ⁴
• What happens if the region is missing or ignored: A viewer that does not read (0018,6011)—or a measurement taken after the frame was exported as a non-DICOM JPEG or uncalibrated Secondary Capture—has only the pixel grid and reports 50 pixels (or 50.0 with no unit). ^{4 6}
• What happens if a custom script mistakes 0003H (cm) for CT/MR millimeters: If software copies 0.05 directly into a millimeter-based spacing variable without multiplying by 10 mm/cm, the same 50-pixel segment is misreported as 2.50 mm (0.25 cm)—an exact 10× underestimation. ^{1 4}

The traps: centimeters, overlapping regions and Doppler areas

Even when Sequence of Ultrasound Regions (0018,6011) is properly populated by the ultrasound cart, four recurring architectural traps cause PACS and workstation measurements to diverge from the console value:

1. The centimeter-versus-millimeter unit trap (0003H = cm): As the Medical Connections technical knowledge base emphasizes ("Crazy as it might seem, DICOM insists on CENTI-metres in this mode"), DICOM PS3.3 Section C.8.5.5.1.15 defines spatial distance in the US Region Calibration Module in centimeters (0003H), centimeter-squared (0008H), or centimeter-cubed (000AH), never millimeters. ^{1 4} Any viewer, research script, or PACS bridge that reads Physical Delta X/Y numerically while assuming DICOM's standard Pixel Spacing (0028,0030) millimeter convention introduces an immediate 10× scale error. ^{1 4}
2. Multi-region defaults and calipers crossing region boundaries: A quad-screen study or dual split-screen comparison legitimately contains multiple sequence items, potentially acquired at different imaging depths (and therefore carrying different Physical Delta X/Y values). Sante DICOM Viewer Pro's documentation illustrates how a region-aware viewer handles this: it outlines calibrated regions on screen (rendering the active region in green and inactive regions in red), exposes a View US Region Details dialog showing the active item's type, units, and delta, lets distance measurements be taken only inside the active region, which the user changes from a right-click menu, and—for area measurements made with a selection tool—defaults calculations to the first calibrated region that has spatial units (cm, mm) unless the user picks another region. ⁵ Those are product-specific choices rather than DICOM mandates. If two split-screen panels were acquired at different depths and a viewer applies Region 1's scale while you measure inside Region 2—or if a caliper crosses the dividing line between two panels—the calculated distance will not match the console. ^{1 5}
3. Overlapping regions and the Region Flags (0018,6016) priority nuance: On color-flow or duplex displays, a color flow or Doppler gate region can overlap the 2D tissue region. PS3.3 Section C.8.5.5.1.3 defines Bit 0 of Region Flags (0018,6016) (0 = high priority, 1 = low priority) to decide which region's pixel component calibration wins where regions intersect, and states plainly: "The lower priority of Region B only applies to its pixel component calibration, not its X and Y direction scaling." ¹ Meanwhile, Bit 1 of Region Flags specifies Scaling Protection (1 = Protected: the region cannot be manually rescaled or overridden by the image reader). ¹ In Philips Lumify R5.0 (HSDP-1299212, Tables 48 and 62), for example, Region Flags (0018,6016) is documented as Always set to 3 (binary 0011, setting both Bit 0 and Bit 1). ⁹
4. Spectral Doppler negative Physical Delta Y and non-spatial axes: In Item 2 of our synthetic example above, the lower spectral Doppler strip has Physical Units X Direction = 0004H (seconds) and Physical Units Y Direction = 0007H (cm/sec). ^{1 9} Because image row coordinates increment downward from top to bottom while positive blood velocities (flow toward the transducer) are plotted upward above the baseline, PS3.3 Sections C.8.5.5.1.16.2 and C.8.5.5.1.17 specify that a negative Physical Delta Y is required so positive velocities point upward. ¹ Furthermore, Reference Pixel x0/y0 (0018,6020)/(0018,6022) is measured relative to Region Location Min x0/y0—not the top-left corner of the whole image. ¹ A naive viewer that attempts a Euclidean distance calculation inside a Doppler or ECG region, or misinterprets a negative Y delta, will produce meaningless results. ^{1 7}

When the sender is at fault: console QA, export settings, recall precedent

Do not assume every measurement disagreement is a PACS viewer problem. The sender side—either the ultrasound export configuration or the scanner's own measurement geometry—is responsible in three distinct scenarios:

1. Export format or secondary-capture stripping: If the ultrasound system or an intermediate capture box exports images as non-DICOM JPEGs, bitmaps, or uncalibrated Secondary Capture objects rather than native Ultrasound Image (1.2.840.10008.5.1.4.1.1.6.1) or Ultrasound Multi-frame Image (1.2.840.10008.5.1.4.1.1.3.1) SOP Instances with (0018,6011) enabled, downstream viewers receive pixel arrays with no region calibration. A February 2024 3D Slicer support thread shows the pattern: a user measuring ultrasound images that had been shared as JPEGs found the ruler unscaled, and a Slicer developer advised obtaining the DICOM files, which store the spacing. ^{2 6} (Only when DICOM transfers fail or drop connections altogether should you branch to hardware transport troubleshooting in our ultrasound DICOM network IO board triage guide—never order a network or I/O board merely because caliper units disagree.)
2. Model- and version-specific module omission: Because the US Region Calibration Module is User Optional (U) in PS3.3 Tables A.6-1 and A.7-1, some older or specific platform releases do not populate (0018,6011) at all—as verified in Toshiba's SSA-640A Viamo V1.20 conformance statement (MIIUS0041EA, Tables 8.1-2 and 8.1-3), which explicitly marks US Region Calibration as Not Present. ^{2 10} Omission can also be conditional on the object: Philips' Lumify R5.0 statement says the module is not present in multi-frame loops that contained a depth or zoom change, even though that system populates regions for other objects. ⁹ On such systems and objects, if measurements must be reviewed downstream, operators must rely on burned-in console calipers or verify whether the system supports coded SR export.
3. On-scanner caliper or acoustic geometry faults (with documented recall precedent): Sometimes the scanner's own on-screen caliper is wrong before the file ever leaves the cart. Our Siemens ACUSON S2000/S3000 board-symptom guide records two terminated FDA recalls of this kind: Z-1832-2011, where any 2D measurement made in Exam Review or offline on a zoomed retrospective clip capture could be erroneous, and Z-2051-2011, a software error affecting measurement results in the patient report. Both were software defects, not board faults. To separate an on-scanner error from an export or viewer error, prove that the console measures known target spacing accurately on a tissue-mimicking phantom using the Gate 3 caliper and geometric-accuracy check in our ultrasound probe acceptance testing protocol, and record the baseline in your routine preventive-maintenance QA log.

When the receiver is at fault: viewer handling and the 2026 PACS recall

When the ultrasound console measures a phantom accurately, burned-in console calipers match the phantom targets, and the exported DICOM header contains a complete Sequence of Ultrasound Regions (0018,6011) with accurate Physical Delta X/Y and 0003H (cm) units, any remaining discrepancy on that same frozen frame points to the receiving side—the viewer itself, or an intermediate system that altered the object it holds. ^{1 3 5}

Primary regulatory evidence from 2026 underscores why clinical engineering and PACS teams must check viewer software revisions whenever ultrasound measurements suddenly disagree. Under FDA Class 2 Device Recall Z-3004-2026 (Recall Event ID 99520, FDA record id=221734, initiated by Merge Healthcare, Inc. on 22 July 2026, classified 21 August 2026, and listed as Open/Classified at review on 15 October 2026), Merge PACS and Merge OrthoPACS medical imaging diagnostic workstations running software version V9.5 (510(k) K192455, 20 units in commerce across CA, FL, KY, NE, and WA) were recalled because a "Software issue may cause measurements to be displayed inaccurately," with the FDA Determined Cause recorded as "Software Design Change." ³ Merge Healthcare's Urgent Field Safety Notice (22 July 2026) asked customers to contact Merge support to identify affected ultrasound studies and set an interim control for its own product—one that also works as a general habit whenever a viewer discrepancy is suspected: "Until the software update is installed, ultrasound measurements performed using the Merge PACS & OrthoPACS Viewer should be confirmed against the measurements acquired at the ultrasound modality." ³

Note the scope boundary of that recall: Z-3004-2026 applies specifically to Merge PACS and Merge OrthoPACS software version V9.5 (20 units in commerce across five states) and is a documented failure mode and interim control, not an incidence statistic or proof that PACS generally mismeasure. ³ What it proves for biomedical and PACS teams is structural: a workstation software update can alter how stored ultrasound measurements are displayed on screen without any change to the ultrasound cart, transducer, or archived DICOM file. ³

How much disagreement is normal: published variability bounds

What if the console measurement and the PACS measurement on a study do not differ by a factor of 10 or report raw pixels, but differ by 1 to 3 millimeters when a radiologist re-measures an organ in PACS? Before declaring either system defective, distinguish re-measurement variability from calibration scale failure. ⁸

A July 2025 open-access clinical study by Alsaadi et al. in the Egyptian Journal of Radiology and Nuclear Medicine (56:114) quantified real-time scanner measurements by two sonographers on a GE Logiq E10 against blinded workstation re-measurements by two radiologists on a GE Centricity PACS workstation (with monthly-calibrated monitors) across 404 abdominal ultrasound images from 300 adult patients (single center, normal anatomy only). ⁸ Even on DICOM-exported studies with high inter-method reliability (intraclass correlation coefficients ICC = 0.886–0.994; Pearson r = 0.888–0.994), blinded re-measurement produced statistically significant mean differences (p < 0.001) with small effect sizes (Cohen's d = 0.06–0.18): ⁸

• Mean scanner-versus-PACS discrepancies by organ: ranged from 0.094 cm (0.94 mm) for the left kidney to 0.269 cm (2.69 mm) for the liver, with real-time ultrasound systematically higher than PACS re-measurements. ⁸
• Bland-Altman 95% limits of agreement: for the liver, individual differences spanned from −1.432 cm to +1.970 cm (and −0.186 cm to +0.374 cm for the left kidney), which the authors attributed to respiratory motion, cursor placement on organ margins, image quality, and display calibration while concluding that measurements were clinically interchangeable in most routine cases. ⁸
• How to use this boundary in engineering triage: Alsaadi et al. measured human cursor re-placement across real-time acquisition and blinded workstation review on one scanner family and one PACS—not software mis-scaling of identical pixel coordinates. ⁸ When different operators measure an organ in real time and again on a workstation, mean differences of about 1–3 mm—with much wider individual limits for the liver—are the documented scale of re-measurement variation. Conversely, when you place a caliper on the exact same pixel endpoints (such as the center of two burned-in console caliper crosses or phantom pin targets) on the same frozen frame, the calibrated distance should agree to within endpoint-placement rounding (about ±1 pixel × Physical Delta per endpoint). Published variability bounds never excuse raw pixel readouts, 10× unit errors, wrong-axis units, or post-update step changes on identical pixel coordinates. ^{1 4 8}

What to document before ordering parts or re-exporting

A cardinal rule of clinical record integrity applies to every calibration investigation: never silently edit or overwrite Sequence of Ultrasound Regions (0018,6011), Physical Delta X/Y, or Pixel Spacing (0028,0030) on archived clinical SOP Instances in PACS to "fix" a viewer readout. The DICOM standard defines what the stored calibration attributes mean; it imposes no display-agreement guarantee and provides no fix-in-place mechanism. ¹ Hand-editing calibration tags on archived clinical records creates unverified scale data, breaks audit lineage between the console and the archive, and can introduce a 10× centimeter-versus-millimeter error across every other viewer that reads the study. ^{1 4}

Instead, preserve the original study on both the modality and the archive, and assemble a concise five-item read-only verification packet before escalating to PACS support, the OEM, or an ultrasound service partner:

1. Modality and probe identity: OEM brand, system model, software/firmware revision, board or transducer part number, and whether the symptom occurs on one probe or across multiple probes.
2. Two-place same-instance measurement log: The SOP Class UID (0008,0016), frame index, landmark pixel coordinates (de-identified, with no patient information), the exact console caliper readout with units, and the exact PACS viewer readout with units. ^{3 5}
3. De-identified (0018,6011) sequence readout: Item count, Region Spatial Format (0018,6012), Region Data Type (0018,6014), Region Flags (0018,6016), pixel bounds (0018,6018)–(0018,601E), Physical Units X/Y (0018,6024)/(0018,6026), and Physical Delta X/Y (0018,602C)/(0018,602E) for each item. ¹
4. Receiver identity and recall check: PACS viewer product name, exact version/build number, confirmation of whether FDA recall Z-3004-2026 (Merge PACS / Merge OrthoPACS V9.5) or another vendor field notice applies, and the measurement result from a second independent region-aware viewer. ^{3 5}
5. Console phantom QA verification (when sender accuracy is in doubt): Known phantom target distance versus console caliper measurement on at least two probes to prove whether the cart's internal acoustic scaling is intact.

Where Rongtao fits—and where it does not

Rongtao Medical is an independent ultrasound parts, board repair, and transducer service organization—not a PACS software vendor, not a third-party viewer integrator, and not a clinical interpretation service. ¹¹ Questions about the clinical significance of any measurement difference on a patient study belong to the responsible clinical team and the facility's quality assurance program. Being explicit about the technical boundary protects both your department and your service budget:

• Where Rongtao does NOT fit: If your two-place verification and header dump show that the ultrasound console measures accurately on a phantom and exports valid Sequence of Ultrasound Regions (0018,6011) attributes—while the discrepancy occurs inside a PACS workstation viewer, a custom script, or a recalled viewer build such as Merge PACS V9.5 (Z-3004-2026)—no ultrasound board or probe replacement will change what PACS displays. ^{1 3} Route that verification packet directly to your PACS administrator and viewer vendor.
• Where Rongtao DOES fit: If a measurement-integrity question arises during post-repair acceptance after a board or probe repair, or if phantom testing shows that the ultrasound system's own on-console scaling, beamformer timing, back-end processor, or transducer array is producing distorted geometry, we support component-level board and probe repair and tested replacement parts. ¹¹ Our documented service process includes photo and video evidence captured at intake, a 48-hour real-machine test of every repaired board inside an actual ultrasound system before release, and QC and packaging records; board repair typically takes 5–8 business days, and the standard warranty is typically 90 days, with final turnaround and warranty terms confirmed per item in the quotation. ¹¹
• Quote-ready evidence handoff: When you contact Rongtao Medical for a verified sender-side or post-repair case, include the OEM, system model, board or probe part number, both measurements with units (console versus PACS/phantom), the de-identified (0018,6011) region-attribute readout, product software versions, quantity, and shipping destination so our engineering desk can scope the repair or replacement accurately.

Frequently asked questions

Our PACS shows a different measurement than the console on the same image—is the scanner broken?
Not necessarily. A stored ultrasound frame does not rely on a single whole-image spacing tag; it uses per-region calibration in Sequence of Ultrasound Regions (0018,6011) under DICOM PS3.3 Section C.8.5.5. ¹ Common causes of a console-versus-PACS disagreement include a PACS viewer ignoring or misapplying region calibration, defaulting to the wrong panel in a multi-region display, a viewer software regression (such as FDA Class 2 recall Z-3004-2026 for Merge PACS and Merge OrthoPACS V9.5), or normal re-measurement variation of a few millimeters. ^{3 5 8} Always measure the exact same pixel landmarks on the console and in PACS and inspect (0018,6011) read-only before suspecting scanner hardware.

Why does our viewer measure some ultrasound images in pixels?
A viewer reports raw pixels when it cannot find or does not support spatial calibration for the region under the cursor. That happens in three main ways: (1) the image was exported as a non-DICOM JPEG or uncalibrated Secondary Capture, which carries no region calibration (as in 3D Slicer support thread 34487, where the user was measuring JPEG copies); ⁶ (2) the scanner model or software version omits the US Region Calibration Module, which is User Optional (U) in PS3.3 Tables A.6-1 and A.7-1 (as verified in Toshiba Viamo V1.20 conformance statement MIIUS0041EA), or omits it for particular objects such as Lumify R5.0 loops with a depth or zoom change; ^{2 9 10} or (3) the caliper endpoints were placed outside the pixel coordinate bounds (0018,6018)–(0018,601E) of any calibrated spatial region, or the viewer only looks for CT/MR Pixel Spacing (0028,0030). ^{1 4}

Do ultrasound DICOM images use Pixel Spacing like CT or MR?
Traditional DICOM Ultrasound Image (A.6) and Ultrasound Multi-frame Image (A.7) objects define calibration per display region via the US Region Calibration Module (PS3.3 Section C.8.5.5), not via a single top-level Pixel Spacing (0028,0030) attribute. ^{1 2} In fact, PS3.3 Section C.8.5.5 contains no mention of Pixel Spacing at all, and its spatial distance unit in Physical Units X/Y Direction (0018,6024)/(0018,6026) is centimeters (0003H = cm) rather than the millimeters used in CT and MR. ^{1 4} Some ultrasound files do carry a top-level Pixel Spacing value as well, as one ITK forum participant notes, but do not assume it matches the calibration of the region under the caliper. ⁷

Can we fix the mismatch by editing the calibration tags in PACS?
No. Never silently edit or inject Sequence of Ultrasound Regions (0018,6011), Physical Delta X/Y (0018,602C)/(0018,602E), or Pixel Spacing (0028,0030) on archived clinical studies. ¹ Modifying calibration tags alters clinical measurement behavior across viewers, breaks audit traceability to the modality, and risks introducing a 10× centimeter-versus-millimeter scaling error. ⁴ Preserve the original instances, confirm measurements on the modality console, and fix the export configuration or viewer software.

What is the 2026 Merge PACS ultrasound measurement recall about?
FDA Class 2 recall Z-3004-2026 (FDA record id=221734, Event ID 99520, initiated 22 July 2026, classified 21 August 2026, status Open) covers Merge PACS and Merge OrthoPACS medical imaging diagnostic workstations running software version V9.5 (20 units in commerce across CA, FL, KY, NE, and WA). ³ Due to a software design change, measurements may be displayed inaccurately. Merge Healthcare's Urgent Field Safety Notice instructs users to confirm ultrasound measurements performed using the Merge PACS & OrthoPACS Viewer against the measurements acquired at the ultrasound modality until the software update is installed. ³

Which calibration applies when a quad image has several regions?
In a multi-region ultrasound image, each panel is encoded as a separate item inside Sequence of Ultrasound Regions (0018,6011) with its own pixel bounding box (Region Location Min/Max x0, y0, x1, y1), Physical Units, and Physical Delta X/Y. ¹ Both caliper endpoints must lie within the single region being measured. Viewers also handle region selection differently—for example, Sante DICOM Viewer Pro takes distance measurements in the active region the user selects, and for selection-tool area measurements defaults to the first calibrated region with spatial units (cm, mm) unless the user picks another. ⁵

How do we verify measurement integrity after a board or probe repair?
First, test on-scanner caliper accuracy directly on the ultrasound console using a calibrated tissue-mimicking phantom across axial and lateral pin targets. Second, freeze and export that phantom frame to PACS, confirm that Sequence of Ultrasound Regions (0018,6011) is populated as declared in the system's DICOM Conformance Statement, and measure the identical landmark pixels on both the console and a region-aware viewer. ^{1 5 9} Comparing the same frozen phantom frame in both places verifies both physical acquisition geometry and DICOM export calibration.