Digital Breast Tomosynthesis QC: Acceptance Testing and Tomosynthesis-Specific Quality Control
By Dr. Troy Zhou
Introduction
Digital breast tomosynthesis (DBT) quality control is the standard MQSA mammography QC program plus a layer of tomosynthesis-specific tests: reconstructed in-plane and z-axis resolution, artifact spread, AEC performance in tomo mode, average glandular dose for tomo and combo acquisitions, geometric sweep accuracy, and ghosting or detector lag, all run under the unit's manufacturer QC manual and verified by an annual medical physicist survey. DBT is regulated as a mammography modality under MQSA, the manufacturer's QC manual is enforceable, and a unit cannot image patients in tomosynthesis mode unless it passes acceptance testing and ongoing QC. 1, 2, 3
Tomosynthesis is the most significant change to mammography since the move from film to full-field digital. Instead of a single projection of a compressed breast, a DBT unit sweeps the x-ray tube through a limited arc, acquires a series of low-dose projection images, and reconstructs a stack of thin in-focus planes. The clinical payoff is the reduction of tissue-overlap masking: structures that would superimpose in a 2D image are separated into different reconstructed slices, which improves cancer detection and reduces false-positive recalls. 3, 4 But the same acquisition that produces those slices introduces new physics to verify, new failure modes to catch, and new dose accounting to perform.
This guide explains how DBT works, how it differs from 2D full-field digital mammography (FFDM) and from synthesized 2D images such as C-View, and the tomosynthesis-specific tests a medical physicist performs at acceptance and annually. It builds on, and does not repeat, our companion guide to general 2D mammography quality control and MQSA physics testing; read that first if you want the foundation, then return here for the tomosynthesis layer. DRPS performs DBT acceptance testing and annual surveys as part of its medical physics consulting and accreditation support services across Florida, Maryland, Virginia, Washington DC, California, and Nevada.
Topic Explanation
How digital breast tomosynthesis works
DBT is a limited-angle tomographic acquisition: the x-ray tube moves through a small arc, typically about 15° to 50° of total sweep depending on the vendor, taking a series of low-dose projection images that are reconstructed into a stack of in-focus planes parallel to the detector. 3, 4 Each reconstructed slice brings one depth into focus and blurs structures at other depths out of the way, so overlapping fibroglandular tissue that could hide or mimic a lesion in a 2D image is partially separated.
Two design parameters dominate the physics:
- Sweep (scan) angle — the total angular range the tube traverses. Wider angles (for example, ~40°–50°) sample the breast from more directions and improve depth discrimination; narrower angles (for example, ~15°) limit depth resolution but reduce sweep time and can favor in-plane sharpness. Vendors deliberately choose different points on this trade-off, which is one reason QC limits are best judged against a unit-specific baseline rather than a universal threshold. 3, 5
- Number and dose of projections — the total sweep dose is divided across many projections, so each projection is low-dose and noisy, and the reconstruction recombines them. This is why a single tomosynthesis acquisition lands at a dose broadly comparable to a single 2D view rather than many times higher. 1, 5
Because the acquisition is limited-angle (not the full 180°+ of CT), DBT cannot achieve true isotropic 3D resolution. In-plane (x–y) resolution is excellent, close to the 2D detector limit, while depth (z) resolution is much coarser and is fundamentally set by the sweep angle. This anisotropy is the defining characteristic of tomosynthesis and drives several of its unique QC tests.
2D FFDM, tomosynthesis, and synthesized 2D (C-View)
Three distinct image types coexist on a modern unit, and the physicist must keep their dose and QC implications straight:
- 2D FFDM — a single full-dose projection of the compressed breast. This is the historical mammogram, and its QC is the established 2D program (AEC, kVp, HVL, MGD, phantom scoring, artifacts, display).
- Tomosynthesis — the limited-angle sweep reconstructed into slices, delivering a per-sweep average glandular dose of its own.
- Synthesized 2D — a 2D-like image (vendor names include C-View and Intelligent 2D) computed from the tomosynthesis projection or reconstruction data. It adds no additional exposure because it reuses the tomosynthesis data. Many sites replaced the separate 2D acquisition with a synthesized image specifically to avoid the added dose of combo mode. 1, 5
This distinction matters for dose accounting. A combo (or combination) acquisition captures both a separate 2D image and a tomosynthesis sweep in the same compression, so the patient receives roughly the sum of the two doses. Replacing the true 2D with a synthesized 2D removes that added exposure while preserving a 2D-style image for the radiologist. Confirming which acquisition modes a unit is actually using clinically is part of a competent dose assessment.
Why DBT needs QC beyond the 2D program
Every 2D requirement still applies to a DBT unit; tomosynthesis QC is additive, not a substitute. 2 The reconstruction step introduces variables the 2D program never measured: the geometric accuracy of the tube sweep, the localization of structures in depth (z), the spread of an artifact or high-contrast object across reconstructed slices, detector lag or ghosting between rapid low-dose projections, and the AEC's behavior when it must set technique for a multi-projection sweep rather than a single shot. These are the tests that distinguish a DBT survey from a 2D one, and they are covered in the unit's QC manual. 2, 3
For the broader multi-modality accreditation picture, see our guide to ACR accreditation physics requirements.
Key Technical Principles
The tomosynthesis-specific test set
The table summarizes the major DBT-specific or DBT-extended tests, what each verifies, and the general nature of the tolerance. As with 2D mammography, the specific numeric limits are defined by the unit's manufacturer QC manual (or the ACR program where applicable), so they are described here generically rather than as universal thresholds. 2, 3, 5
| Test | What it checks | Tolerance / action level |
|---|---|---|
| Reconstructed in-plane spatial resolution / MTF | Sharpness within a reconstructed slice (x–y) | At or above the manual's baseline; in-plane resolution should approach the 2D detector limit 2 |
| Z-axis (depth) resolution / artifact spread function (ASF) | How tightly a high-contrast object is confined in depth across slices | ASF width (full-width at some fraction of maximum) within the baseline range; narrower means better z-localization 5, 7 |
| Z-localization / geometric sweep accuracy | A test object reconstructs at its true height and position | Reconstructed depth and in-plane position within the manual's geometric tolerance 2 |
| AEC performance and reproducibility (tomo mode) | Consistent detector signal and technique selection across the sweep and across thicknesses | AEC output/signal and reproducibility within the manual's tolerance for tomo mode 2 |
| Average glandular dose (AGD/MGD), tomo and combo | Dose to the standard phantom per sweep and per combo acquisition | Per-sweep AGD within the manual's limit; combo mode evaluated as the summed dose 1, 6 |
| Ghosting / detector lag | Residual signal from one projection contaminating the next | No clinically significant ghosting; residual within the manual's limit 2 |
| Reconstructed-slice artifact evaluation | Reconstruction, detector, or motion artifacts in the slice stack | No clinically significant artifacts; investigate and correct any found 2 |
| Missed-tissue at the chest wall | Tissue near the chest wall reconstructed and included | Missed tissue within the manual's allowance (extends the 2D chest-wall coverage check) 2 |
| Phantom image quality (DBT) | Object detection in the reconstructed slices of the applicable phantom | Minimum object scores per the unit's QC manual / ACR DBT requirements 2, 3 |
The artifact spread function (ASF) deserves emphasis because it is the practical surrogate for z-axis resolution: a small, high-contrast object should appear sharp in one reconstructed plane and fade quickly in adjacent planes. A wide ASF means the object's signal smears across many slices, which is exactly the out-of-plane blur a wider sweep angle is meant to suppress. 5, 7
The display side still matters too. The radiologist scrolls through a reconstructed slice stack on a review workstation, so monitor luminance, uniformity, and grayscale calibration remain part of the imaging chain, as covered in our note on the SMPTE pattern for monitor QC. Detector signal behavior also connects to the broader idea of an exposure index in digital radiography, where consistent detector response underpins dose-quality control.
Scan angle and z-axis resolution
The single most important geometric principle in tomosynthesis is that depth (z) resolution improves with a wider scan angle, while a narrow angle limits z-resolution. Conceptually, reconstructing depth requires viewing a structure from a range of directions: the more angular spread the sweep provides, the better the reconstruction can triangulate where a structure lives along the depth axis. With only a limited arc, structures remain blurred along z, and the blur grows as the available angle shrinks.
A simple way to express the limited-angle blur idea is to relate the in-depth blur of a reconstructed point to the sweep half-angle. For a feature at height
so a wider
Average glandular dose: the Dance model extended for tomosynthesis
Average glandular dose (AGD, equivalent to mean glandular dose, MGD) is not measured directly in the breast. The physicist measures incident air kerma at the phantom surface for the technique the unit selects, then converts it to glandular dose with published coefficients. For 2D mammography the standard is the Dance formulation: 6
where:
is the incident air kerma at the upper surface of the breast or standard phantom (no backscatter), for the clinical technique; converts incident air kerma to AGD for a standard-glandularity breast, indexed by HVL and compressed thickness; corrects for breast composition (glandularity) differing from the reference; and corrects for the target/filter spectrum relative to the reference. For tomosynthesis, the relevant spectrum/geometry correction must reflect the tube-motion sweep, not a single static projection. 6, 7
For DBT, Dance and colleagues extended this model with tomosynthesis-specific factors: a
where
Worked example: per-projection, per-sweep, and combo AGD
Suppose a DBT unit images the standard phantom with a 15-projection sweep. The physicist measures the total incident air kerma for the whole sweep as
Spread across 15 projections, that is roughly
Now suppose the same patient is imaged in combo mode, adding a separate 2D acquisition whose AGD computes (via
which now sits near the MQSA 3.0 mGy reference for a single standard-phantom view, illustrating exactly why combo mode draws dose scrutiny and why a synthesized 2D (which adds no exposure) is the lower-dose alternative to a true combo acquisition. 1, 6 These figures are illustrative; in practice the physicist reads
Clinical Impact
Better detection, but new physics to police
The clinical case for DBT is strong: separating overlapping tissue improves cancer detection and reduces false-positive recalls compared with 2D mammography alone. 3, 4 But every benefit rides on the reconstruction being geometrically correct and the dose being properly accounted for. A unit with a drifting sweep geometry can mislocalize a lesion in depth; a unit with excessive artifact spread can smear a microcalcification cluster across slices until it loses conspicuity; a unit run carelessly in combo mode can deliver roughly twice the per-view dose of a synthesized-2D workflow. DBT-specific QC is what keeps the detection benefit from being undermined by a quiet technical fault.
Where DBT QC catches problems 2D QC would miss
Several failure modes are invisible to a 2D-only program. A geometric calibration error in the tube sweep degrades z-localization without necessarily changing the 2D image. Detector lag between the rapid low-dose projections produces ghosting that only appears in the reconstructed stack. A reconstruction-filter change after a software update can shift the artifact spread function and the noise texture while leaving 2D phantom scores untouched. Based on articles retrieved from PubMed, the harmonized QC program for the multi-site TMIST tomosynthesis trial found that the largest sources of QC non-conformance were operator error, deviations from protocol, and unreported software updates and preventive-maintenance activities that shifted QC setpoints, and that drops in signal-difference-to-noise ratio or rising mAs sometimes preceded tube failure. 5 The practical lesson is that DBT QC must be baseline-relative and must be re-baselined after any hardware or software change.
The synthesized-2D dose decision is a QC and safety question
Because synthesized 2D adds no dose while a true combo adds a full 2D exposure, the choice between them is a measurable patient-safety decision, not just a workflow preference. The physicist's dose report should make the per-mode AGD explicit so the facility can see what combo imaging actually costs each patient and confirm that any continued use of true combo mode is clinically justified rather than a default left in place at install. 1, 5, 6
Practical Optimization Tips
A defensible DBT QC program runs the 2D program faithfully and then layers the tomosynthesis tests on top, on a disciplined cadence, against unit-specific baselines.
1. Use the unit's manufacturer QC manual, and re-baseline after changes
Each DBT unit operates under its manufacturer's FDA-accepted QC manual, which defines the tomosynthesis tests, frequencies, and tolerances; these differ meaningfully between vendors because the acquisition geometry and reconstruction differ. 2, 3 Establish baselines at acceptance, and re-baseline after any tube replacement, detector service, software/reconstruction update, or relocation. Unreported updates that silently shift setpoints are a documented top cause of QC drift. 5
2. Track z-resolution and artifact spread as trends
A single passing ASF or z-localization value can hide slow geometric drift. Trend the artifact spread function width and z-localization over time; a widening ASF often signals a developing geometric or reconstruction problem before it is obvious clinically. 5, 7
3. Make per-mode dose explicit
Report AGD separately for tomo, 2D, and combo acquisitions, and confirm which modes are in clinical use. If a site is running true combo by default, quantify the added dose and confirm it is intended; synthesized 2D is the no-added-dose alternative. 1, 6
4. Verify AEC behavior in tomo mode specifically
The AEC must set technique sensibly for a multi-projection sweep across the clinical thickness range, not just for a single 2D shot. Check AEC output and reproducibility in tomo mode across phantom thicknesses, because tomo-mode AEC logic can differ from 2D-mode logic. 2
5. Keep the slice-review chain in the program
The diagnosis is made by scrolling reconstructed slices, so the review workstation's luminance, uniformity, and grayscale calibration are part of DBT QC, not an afterthought. 2
Common pitfalls to avoid
- Treating DBT as covered by 2D QC. Tomosynthesis adds z-resolution, artifact spread, geometric, ghosting, and per-mode dose tests that 2D QC never touches.
- Forgetting to re-baseline after a software update. Reconstruction changes can move QC setpoints invisibly; re-baseline and document.
- Leaving true combo mode on by default. It can roughly double per-view dose versus synthesized 2D; make the choice deliberate and documented.
- Ignoring chest-wall missed tissue in 3D. Missed-tissue checks extend to the reconstructed volume, not just the 2D field.
- Applying universal thresholds across vendors. Different sweep angles and reconstruction filters mean limits must be baseline-relative. 5
Regulatory Considerations
Digital breast tomosynthesis is regulated as a mammography modality under MQSA (21 CFR Part 900): the FDA approved DBT systems for clinical use, each unit operates under the manufacturer's FDA-accepted QC manual, and that manual is enforceable under MQSA. 1, 2 The annual medical physicist survey must include the tomosynthesis tests in that manual, and ACR accreditation carries DBT-specific requirements in addition to 2D accreditation. 3
Key frameworks to reference:
- MQSA, 21 CFR Part 900 — the federal regulation governing mammography accreditation, certification, personnel, equipment, QC, and the mean glandular dose framework; DBT falls under it as an approved modality. 1
- FDA DBT-under-MQSA policy guidance — the FDA's position that tomosynthesis units are imaged under their FDA-accepted QC manuals, which become the enforceable QC program for the unit. 2
- FDA 21 CFR 1020.30 — the federal performance standard for diagnostic x-ray systems (beam quality, reproducibility, collimation) that the mammography/DBT survey also touches. 8
- ACR Mammography Accreditation Program — the most common FDA-approved accreditation route, which includes DBT-specific phantom and clinical-image requirements for tomosynthesis units. 3
- ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of Mammography Equipment — the professional standard describing the qualified medical physicist's role and survey scope, including tomosynthesis. 9
- Dance tomosynthesis dosimetry formalism — the t-/T-factor extension of the Dance AGD model that underlies tomosynthesis dose estimation in the UK, European, and IAEA protocols. 7 Jurisdiction note. DBT units are x-ray machines, so they fall under FDA and state radiation-control authority rather than the NRC, with MQSA layering a federal certification scheme on top. Of the states DRPS serves, Florida, Maryland, Virginia, California, and Nevada administer their own radiation-control programs, while Washington, DC is within the NRC's direct authority for byproduct material; in all of them, mammography and DBT certification flow through MQSA and an FDA-approved accreditation body. A facility adding tomosynthesis should confirm its MQSA certification covers the DBT unit, that the unit's QC manual is on file, and that the annual physicist survey includes the tomosynthesis tests. The acceptance/equipment-evaluation report, annual survey, and routine QC logs should be retained as a coherent, inspection-ready package, consistent with the FDA's EQUIP emphasis on whether facilities act on QC findings. 1, 3 DRPS aligns DBT acceptance testing, the annual survey, documentation, and accreditation submission as part of medical physics consulting and accreditation support.
Frequently Asked Questions (FAQs)
What is digital breast tomosynthesis and how is it different from 2D mammography?
Digital breast tomosynthesis (DBT) is a limited-angle acquisition in which the x-ray tube sweeps through a small arc, typically about 15 to 50 degrees depending on the vendor, taking a series of low-dose projection images. Those projections are reconstructed into a stack of thin in-focus planes that reduce the masking effect of overlapping tissue. Standard 2D full-field digital mammography (FFDM) instead captures a single projection of the whole compressed breast, so overlapping structures can hide or mimic a lesion.
Does digital breast tomosynthesis require its own quality control?
Yes. DBT adds tomosynthesis-specific QC on top of the standard 2D mammography program, including reconstructed in-plane and z-axis (depth) resolution, artifact spread, in-plane spatial resolution or MTF, AEC performance and reproducibility in tomo mode, average glandular dose for tomo and combo modes, geometric and sweep accuracy, and ghosting or detector lag. These tests are defined by the unit's QC manual and do not replace any 2D requirement.
How does tomosynthesis affect radiation dose compared to 2D mammography?
A single tomosynthesis acquisition delivers a dose broadly comparable to a single 2D view, because the total sweep dose is split across many low-dose projections. The dose concern arises in combo mode, where a 2D image and a tomosynthesis sweep are both acquired in the same compression, roughly adding the two doses. Synthesized 2D images such as C-View are reconstructed from the tomosynthesis data and add no extra exposure, which is one reason many sites replaced the separate 2D acquisition with a synthesized image.
How is average glandular dose calculated for tomosynthesis?
Average glandular dose for DBT extends the Dance conversion-coefficient model used for 2D mammography. The physicist measures incident air kerma and applies tabulated g, c, and s factors, then a tomosynthesis t-factor for a single projection or a T-factor for the full sweep that accounts for the tube-motion geometry. The result is the per-sweep average glandular dose, and in combo mode the tomosynthesis and 2D doses are summed.
Why does scan angle matter for tomosynthesis image quality?
The width of the tube sweep angle controls depth (z-axis) resolution. A wider scan angle samples the breast from more directions and improves the ability to separate structures at different depths, while a narrow angle limits z-resolution and leaves more out-of-plane blur. Wider angles generally improve mass and margin visibility, while narrower angles tend to favor in-plane sharpness and microcalcification rendering, so vendors choose different trade-offs.
Is digital breast tomosynthesis regulated under MQSA?
Yes. The FDA approved DBT as a mammography modality under the Mammography Quality Standards Act, and facilities operate each DBT unit under the manufacturer's FDA-accepted QC manual, which is enforceable under MQSA. The annual medical physicist survey must include the tomosynthesis tests in that manual, and ACR accreditation includes DBT-specific requirements. A facility cannot legally image patients in tomosynthesis mode without meeting these standards.
Key Takeaways
- DBT QC is the 2D program plus a tomosynthesis layer. Reconstructed in-plane and z-axis resolution, artifact spread, tomo-mode AEC, per-mode AGD, geometric sweep accuracy, and ghosting are added on top of every existing 2D requirement.
- Scan angle sets depth resolution. A wider sweep angle improves z-axis (depth) localization; a narrow angle limits it, which is why z-resolution and artifact spread are baseline-relative metrics rather than universal thresholds.
- Dose accounting is mode-specific. A single tomo sweep is comparable to a single 2D view, true combo mode roughly adds the two doses, and synthesized 2D (C-View) adds no exposure.
- The Dance model extends to tomosynthesis. AGD uses
with an added per-sweep -factor that accounts for the tube-motion geometry. - Baselines must be re-established after changes. Software, reconstruction, tube, or detector changes can silently shift QC setpoints; re-baseline and document after any of them.
- DBT is enforceable under MQSA. The manufacturer's QC manual is the enforceable program, the annual physicist survey must include the tomo tests, and ACR accreditation has DBT-specific requirements.
Conclusion
Digital breast tomosynthesis delivers a real clinical gain by separating overlapping tissue, but it earns that gain only when the reconstruction is geometrically faithful, the depth resolution holds to baseline, and the dose is honestly accounted for across tomo, 2D, and combo modes. The tomosynthesis-specific tests, reconstructed in-plane and z-axis resolution, artifact spread, geometric sweep accuracy, tomo-mode AEC, ghosting, and per-mode AGD, are what turn that requirement into verifiable physics.
A strong DBT program runs the full 2D mammography QC faithfully, layers the tomosynthesis tests on top against unit-specific baselines, re-baselines after every hardware or software change, and reports per-mode dose so the facility can make the synthesized-versus-combo decision deliberately. Under MQSA, the manufacturer's QC manual is the enforceable program and the annual medical physicist survey is the backbone: it verifies the limited-angle physics, anchors the documentation, and gives the facility defensible evidence that every tomosynthesis study is acquired safely and read on a calibrated chain.
How DRPS Can Help
Diagnostic Radiation Physics Services performs DBT acceptance testing and mammography equipment evaluations, runs the annual MQSA medical physicist survey including the tomosynthesis-specific tests, measures per-mode average glandular dose, establishes and re-baselines z-resolution and artifact-spread metrics, reviews technologist QC, and prepares facilities for ACR accreditation and FDA inspection of tomosynthesis units. This work fits within medical physics consulting and accreditation support.
DRPS supports facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, and Nevada. If you are adding a DBT unit, switching to a synthesized-2D workflow, or preparing for accreditation or an EQUIP inspection of a tomosynthesis system, contact DRPS to schedule acceptance testing or the annual survey.
A well-run tomosynthesis QC program makes the depth-resolved, low-dose, geometrically faithful image the routine result, not the lucky one.
Related Resources
- Mammography quality control and MQSA physics testing
- ACR accreditation physics requirements
- Exposure index in digital radiography
- SMPTE pattern for monitor QC
- Accreditation support
- Medical physicist consulting
- Contact DRPS
References
- U.S. Food and Drug Administration. Mammography Quality Standards Act (MQSA); 21 CFR Part 900. ecfr.gov
- U.S. Food and Drug Administration. Digital Accreditation and Digital Breast Tomosynthesis under MQSA (MQSA Insights). fda.gov
- American College of Radiology. Mammography Accreditation Program Requirements (including Digital Breast Tomosynthesis). acr.org
- U.S. Food and Drug Administration. Breast Tomosynthesis (3D Mammography): What Patients and Providers Should Know. fda.gov
- Maki AK, Mawdsley GE, Mainprize JG, et al. Quality control for digital tomosynthesis in the ECOG-ACRIN EA1151 TMIST trial. Med Phys. 2023;50(12):7441-7461. doi:10.1002/mp.16786. doi.org
- Dance DR, Skinner CL, Young KC, Beckett JR, Kotre CJ. Additional factors for the estimation of mean glandular breast dose using the UK mammography dosimetry protocol. Phys Med Biol. 2000;45(11):3225-3240. doi:10.1088/0031-9155/45/11/308. doi.org
- Dance DR, Young KC, van Engen RE. Estimation of mean glandular dose for breast tomosynthesis: factors for use with the UK, European and IAEA breast dosimetry protocols. Phys Med Biol. 2011;56(2):453-471. doi:10.1088/0031-9155/56/2/011. doi.org
- U.S. Food and Drug Administration. 21 CFR 1020.30: Diagnostic X-ray Systems and Their Major Components. ecfr.gov
- American College of Radiology and American Association of Physicists in Medicine. ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of Mammography Equipment. acr.org
- American Association of Physicists in Medicine. AAPM reports and task-group resources on breast tomosynthesis and breast dosimetry. aapm.org
- Bushberg JT, Seibert JA, Leidholdt EM, Boone JM. The Essential Physics of Medical Imaging. 4th ed. Wolters Kluwer; 2021. wolterskluwer.com