Synthesized 2D Mammography: Physics & QC

Synthesized 2D mammography reconstructs a conventional-looking two-dimensional image directly from the tomosynthesis projection data, so a combined screening exam no longer needs a separate full-field digital mammography (FFDM) exposure. That single change roughly halves the radiation dose of combo-mode screening while preserving cancer detection — but the synthesis algorithm is a new link in the imaging chain, and it must be validated and quality-controlled like any other.

Introduction

When digital breast tomosynthesis (DBT) first entered screening, it was used in "combo mode": the system acquired a standard FFDM image and then a tomosynthesis sweep, giving the reader both a familiar 2D overview and the 3D slices. The clinical benefit was real, but so was the cost — two acquisitions meant roughly double the radiation dose of 2D alone, and each acquisition counts against the mammography dose limit.

Synthesized 2D mammography was developed to remove that penalty. Instead of taking a second exposure for the 2D image, the system computes a 2D-style image from the tomosynthesis data it already has. The U.S. Food and Drug Administration first approved a synthesized 2D mode (Hologic's C-View) as a supplement to the Selenia Dimensions tomosynthesis premarket approval in 2013, allowing the synthesized image to be used in place of the conventional 2D exposure in a tomosynthesis screening exam ⁵.

This article explains the physics of how a 2D image is synthesized from 3D data, why the dose savings are what they are, what the clinical evidence shows, and — most importantly for a facility — what a medical physicist must verify before and after a synthesized-2D system goes live. DRPS provides this review as part of its mammography physics and MQSA and accreditation support services across Florida, Maryland, Virginia, Washington DC, California, Nevada, Pennsylvania, New York, New Jersey, and Delaware.

Topic Explanation

What "synthesized 2D" actually means

A synthesized 2D image is a two-dimensional mammogram computed from the tomosynthesis slice stack, not acquired with its own x-ray beam. The patient is compressed once; the system performs the tomosynthesis sweep; the reconstruction produces a set of thin slices through the breast; and a synthesis algorithm then collapses those slices into a single 2D image that resembles a conventional mammogram.

The appeal is straightforward. In a combined (combo-mode) exam, the separate FFDM exposure exists only to give the reader a familiar 2D layout and to make year-to-year comparison easier. If that 2D layout can be generated from data already acquired, the separate exposure — and its dose — can be eliminated. For the DBT acquisition itself, see our companion post on digital breast tomosynthesis QC.

Importantly, synthesized 2D is validated as a replacement for the separate 2D exposure within a combined exam. It is interpreted together with the tomosynthesis slices from which it was created, not as a standalone mammogram.

Why it is not just a "flattened" tomosynthesis image

Naively, one could average the slice stack into a single image, but that would blur exactly the features radiologists rely on. The early synthesis approach instead used a maximum-intensity-projection (MIP) style operation that preserves and emphasizes high-attenuation structures — microcalcifications and spiculated margins — as the stack is collapsed to 2D ⁵. Later-generation algorithms use feature detection and blending so that calcifications, masses, and tissue background are each rendered appropriately rather than a single projection operator being applied everywhere.

The result looks like a mammogram but is not identical to FFDM: contrast and texture differ, and readers must learn its appearance. This is a key physics-and-practice point — the synthesized image is a computed representation, and the algorithm's behavior is part of the imaging chain's performance.

Key Technical Principles

The dose arithmetic of combo mode

The dose case for synthesized 2D is the clearest part of the physics. Consider the per-view mean glandular dose (MGD) in a combined exam. The combined (FFDM + DBT) per-view dose is the sum of the two acquisitions:

D_{combo} = D_{FFDM} + D_{DBT}

When the FFDM exposure is replaced with a synthesized image, which requires no radiation, the per-view dose becomes the tomosynthesis acquisition alone:

D_{synth+DBT} = D_{s2D} + D_{DBT} = 0 + D_{DBT} = D_{DBT}

The fractional dose reduction is therefore the FFDM share of the combo dose:

Reduction = \frac{D _{FFDM}}{D _{combo}} = \frac{D _{FFDM}}{D _{FFDM} + D _{DBT}}

A worked example

A per-view dose analysis of a screening cohort reported a mean glandular dose for a single craniocaudal view of about 1.37 mGy for FFDM and about 1.86 mGy for DBT, with DBT delivering on average about 38% more dose per view than FFDM ⁴. Using those values, the combo-mode per-view dose is:

D_{combo} = 1.37 + 1.86 = 3.23 mGy

Replacing the FFDM exposure with a synthesized image leaves only the DBT dose of about 1.86 mGy, so the per-view reduction is:

Reduction = \frac{1.37}{3.23} \approx 0.42 = 42%

In other words, synthesized 2D removes roughly 40 to 50 percent of the per-view dose of a combined exam, with the exact figure depending on the relative FFDM and DBT doses on a given system and breast. Critically, each acquisition in a combined exam remains subject to the MQSA limit on mean glandular dose, so eliminating one acquisition directly reduces how much of that budget the exam consumes ⁶.

Per-view dose and the MQSA limit

Acquisition	Representative per-view MGD (CC view)	Radiation exposure?
FFDM (2D) alone	~1.37 mGy	Yes
DBT (tomosynthesis sweep)	~1.86 mGy	Yes
Combo: FFDM + DBT	~3.23 mGy	Two exposures
Synthesized 2D (from DBT data)	0 mGy (computed)	No
Synthesized 2D + DBT	~1.86 mGy	One exposure
MQSA regulatory limit (standard phantom, per view)	3.0 mGy	—

Representative per-view doses are from a per-view screening analysis and will vary with breast thickness, density, and system ⁴; the 3.0 mGy figure is the MQSA regulatory limit for a standard-phantom craniocaudal view ⁶. The table makes the point visually: combo mode can approach the regulatory per-view limit, while synthesized 2D plus DBT sits comfortably below it. For the dosimetry of the 2D exposure itself, see mean glandular dose in mammography.

Clinical Impact

Does removing the FFDM exposure cost detection?

The central clinical question is whether a computed 2D image performs as well as an acquired one. The evidence accumulated over several years says that, in a combined exam, it does.

In the Oslo prospective screening trial of 24,901 examinations, reconstructed 2D images plus DBT performed comparably to FFDM plus DBT, with cancer detection rates of 7.8 versus 7.7 per 1000 in the period using the current-generation synthesis, and the authors concluded the combination was adequate for routine clinical use ¹.
In a fully crossed multireader study, synthesized mammography alone achieved a mean area under the ROC curve of 0.894 versus 0.889 for FFDM alone, and synthesized-plus-DBT was comparable to FFDM-plus-DBT, leading the authors to conclude synthesized 2D may eliminate the need for FFDM as part of a routine study ².
In a large community-based practice of 78,810 screening mammograms, DBT with synthesized 2D produced a significantly lower recall rate (4.3%) than DBT with FFDM (5.8%) and FFDM alone (8.7%), with a comparable cancer detection rate and a higher positive predictive value ³.

A representative comparison

Metric	FFDM + DBT (combo)	Synthesized 2D + DBT
Per-view radiation dose	Higher (two acquisitions)	Lower (one acquisition)
Cancer detection rate	Reference	Comparable ^{1, 3}
Recall rate	Reference	Comparable or lower ³
Positive predictive value	Reference	Comparable or higher ³
2D image origin	Acquired exposure	Computed from DBT data
Reader familiarity	Native mammographic appearance	Requires learning a new appearance

The practical message is that synthesized 2D is a dose-reduction technology that does not require sacrificing the 2D overview or detection performance — provided the synthesis is working correctly, which is where quality control comes in.

The reader-training caveat

Because the synthesized image differs in appearance from FFDM, interpretation is not automatically transferable. Facilities adopting synthesized 2D should expect a learning curve, particularly for microcalcification conspicuity, and should treat reader familiarity as part of a successful implementation rather than an afterthought.

Practical Optimization Tips

For the medical physicist, synthesized 2D adds a computed step to the chain that must be verified. The acquisition QC does not disappear; the synthesis QC is added on top.

1. Confirm the regulatory pathway and QC manual

Synthesized 2D operates under the facility's tomosynthesis approval and the FDA's framework for DBT. Under MQSA, the facility must run QC according to the manufacturer's manual for the specific system, and the synthesized image is part of that program ⁶. Confirm the correct, current manual is in use.

2. Verify calcification and mass rendering

Because early synthesis emphasizes high-attenuation features, the physicist should confirm that microcalcifications and spiculated margins are faithfully represented in the synthesized image and are not lost, duplicated, or exaggerated by the algorithm. This is best checked with the manufacturer's phantom and the ACR Digital Mammography QC phantom, correlating synthesized appearance with the slice data. See ACR digital mammography phantom QC.

3. Hunt for synthesis artifacts

Collapsing a 3D stack into 2D can introduce artifacts — out-of-plane blur mapped into the plane, overlapping-structure artifacts, or bright-line artifacts around high-contrast objects. Review the synthesized image for structures that do not correspond to real anatomy on the slices.

4. Keep the dose and phantom image-quality tests

Replacing FFDM with a synthesized image changes the exposures delivered, but the DBT acquisition still must pass the mean glandular dose limit and the phantom image-quality requirements. Do not let the dose-reduction benefit distract from verifying that the remaining acquisition is within limits ⁶. For the broader MQSA test set, see mammography QC under MQSA.

5. Validate after any software update

The synthesis algorithm is software. A version change can alter calcification rendering or introduce new artifacts without any hardware change. Treat a synthesis-software upgrade as a trigger for re-verification, the same way you would treat a reconstruction change on CT or MRI.

Common pitfalls to avoid

Treating synthesized 2D as "free." It is dose-free, but it is not QC-free; it adds a verification step.
Assuming FFDM-equivalent appearance. The image is computed and looks different; reader training matters.
Using it standalone. Synthesized 2D is validated with the DBT slices, not as a lone mammogram.
Skipping re-verification after updates. Algorithm changes can silently change image content.
Ignoring the remaining dose test. The DBT acquisition still must meet the MQSA mean glandular dose limit.

Regulatory Considerations

Synthesized 2D mammography is regulated under MQSA, operates under the tomosynthesis system's FDA approval, and is quality-controlled under the manufacturer's QC manual. It is not a separate modality with its own standalone accreditation; it is a mode of a tomosynthesis unit.

Key regulatory anchors:

MQSA, 21 CFR Part 900 — the federal mammography quality program. It sets the mean glandular dose limit of 3.0 mGy per exposure for a standard-phantom craniocaudal view, the phantom image-quality requirements, personnel qualifications, and the annual medical physicist survey ⁶.
FDA PMA P080003 and supplement S001 — the premarket approval of the Selenia Dimensions tomosynthesis system and the C-View synthesized 2D software, which permits the synthesized image to replace the conventional 2D exposure in a tomosynthesis screening exam ⁵.
FDA framework for DBT — tomosynthesis systems and their synthesized-2D options are approved devices operated under MQSA, and facilities must follow the manufacturer's QC manual as the applicable quality-control standard ⁶.

Because x-ray mammography units are FDA- and state-regulated (not NRC-regulated byproduct material), a facility must also satisfy its state radiation-control program's registration and inspection requirements for the x-ray unit. Of the states DRPS serves, all license or register x-ray machines through their state programs in addition to the federal MQSA requirements. The annual medical physicist survey required under MQSA is the natural point at which to verify the synthesized-2D mode, document its performance, and keep the facility accreditation-ready. DRPS supports this through mammography physics and MQSA and accreditation support.

Frequently Asked Questions (FAQs)

What is synthesized 2D mammography?

Synthesized 2D mammography is a conventional-looking two-dimensional image reconstructed from the digital breast tomosynthesis (DBT) projection data rather than acquired with its own x-ray exposure. It lets a combined screening exam provide both a 2D overview image and the DBT slices while taking only the tomosynthesis exposure, eliminating the separate full-field digital mammography (FFDM) acquisition that combo mode otherwise requires.

Does synthesized 2D reduce radiation dose?

Yes. In combined screening, the synthesized image carries no additional radiation because it is computed from the tomosynthesis data already acquired. Replacing the separate FFDM exposure with a synthesized image removes roughly one of the two acquisitions in combo mode, which can cut the per-view dose of a combined exam by about 40 to 50 percent while keeping the tomosynthesis information.

Is synthesized 2D as good as full-field digital mammography?

Large screening and reader studies have found that synthesized 2D combined with tomosynthesis performs comparably to full-field digital mammography combined with tomosynthesis for cancer detection, and some studies report lower recall rates and higher positive predictive value. The synthesized image is now considered adequate for routine clinical use in a combined exam, though the image has a different appearance that readers must learn.

How is the synthesized image created?

The tomosynthesis reconstruction produces a stack of thin slices through the breast. The synthesis algorithm collapses that stack into a single 2D image, historically by a form of maximum-intensity projection that emphasizes high-attenuation features such as calcifications and spiculated margins, with later versions using feature-based blending. The physics goal is to present suspicious findings on a familiar 2D layout without a second exposure.

Does synthesized 2D need its own quality control?

Yes. Under MQSA, a facility must perform quality control according to the manufacturer's manual for its tomosynthesis system, and the synthesized image is part of that chain. The medical physicist should confirm that calcifications and masses are faithfully represented, that artifacts are controlled, and that the full acquisition still meets the mean glandular dose limit and phantom image-quality requirements.

Can synthesized 2D be used as a standalone screening mammogram?

No. Synthesized 2D is approved and validated as a replacement for the separate 2D exposure within a combined tomosynthesis exam, not as a standalone mammogram without the DBT slices. The synthesized image is interpreted together with the tomosynthesis dataset from which it was created.

What dose limit applies to a mammography exposure?

Under MQSA, the mean glandular dose for a single craniocaudal view of a standard phantom representing an average breast must not exceed 3.0 mGy per exposure. Each acquisition in a combined exam is subject to this limit, which is one reason removing the separate FFDM exposure with a synthesized image is attractive for dose management.

Key Takeaways

Synthesized 2D is computed, not acquired. It reconstructs a 2D-style image from the tomosynthesis slice stack, so it adds no radiation dose.
It roughly halves combo-mode dose. Removing the separate FFDM exposure cuts the per-view dose of a combined exam by about 40 to 50 percent.
Detection is preserved. Screening and reader studies show synthesized-2D-plus-DBT is comparable to FFDM-plus-DBT, sometimes with lower recall and higher PPV.
The appearance differs. The synthesized image is not identical to FFDM; readers need to learn it.
It is a QC item, not a free lunch. The synthesis algorithm must be verified for calcification rendering and artifacts, and the DBT acquisition still must meet MQSA dose and phantom limits.
Software updates trigger re-verification. Synthesis is software, and a version change can alter image content.

Conclusion

Synthesized 2D mammography is a clean example of using computation to remove a radiation cost. By generating the 2D overview from tomosynthesis data already acquired, it eliminates an entire exposure from combined screening, cutting per-view dose substantially while preserving the diagnostic performance that made combo mode attractive in the first place. The evidence from large screening cohorts and reader studies supports its routine use, and its dose advantage is simple arithmetic.

What the technology does not eliminate is the physicist's responsibility. The synthesis algorithm is a new, software-defined link in the imaging chain, and like reconstruction on any modality it can fail silently — losing a calcification, introducing an artifact, or changing behavior after an update. A facility that treats synthesized 2D as a validated, QC'd component rather than a free convenience gets the dose benefit without surrendering image quality.

How DRPS Can Help

Diagnostic Radiation Physics Services helps mammography facilities adopt and maintain tomosynthesis and synthesized-2D imaging within MQSA. This can include the annual medical physicist survey, verification of synthesized-image quality and artifact behavior, mean glandular dose and phantom image-quality testing, QC-program setup against the manufacturer's manual, and keeping the facility accreditation-ready through mammography physics and MQSA and accreditation support.

DRPS supports facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware. If your facility is enabling synthesized 2D or preparing for an MQSA inspection, contact us to review the mode before it goes live.

Related Resources

References

Skaane P, Bandos AI, Eben EB, et al. Two-view digital breast tomosynthesis screening with synthetically reconstructed projection images: comparison with digital breast tomosynthesis with full-field digital mammographic images. Radiology. 2014;271(3):655-663. doi:10.1148/radiol.13131391. PubMed
Zuley ML, Guo B, Catullo VJ, et al. Comparison of two-dimensional synthesized mammograms versus original digital mammograms alone and in combination with tomosynthesis images. Radiology. 2014;271(3):664-671. doi:10.1148/radiol.13131530. PubMed
Aujero MP, Gavenonis SC, Benjamin R, Zhang Z, Holt JS. Clinical Performance of Synthesized Two-dimensional Mammography Combined with Tomosynthesis in a Large Screening Population. Radiology. 2017;283(1):70-76. doi:10.1148/radiol.2017162674. PubMed
Gennaro G, Bernardi D, Houssami N. Radiation dose with digital breast tomosynthesis compared to digital mammography: per-view analysis. Eur Radiol. 2018;28(2):573-581. doi:10.1007/s00330-017-5024-4. PubMed
U.S. Food and Drug Administration. Premarket Approval (PMA) P080003/S001 — Selenia Dimensions 3D System with C-View software. 2013. accessdata.fda.gov
U.S. Food and Drug Administration. Mammography Quality Standards Act — 21 CFR Part 900. ecfr.gov