Mammography Quality Standards Act (MQSA) & Mammography Physics

Mammography Quality Standards Act (MQSA) Framework

Enacted in 1992 and implemented in 21 CFR Part 900, the Mammography Quality Standards Act gives the FDA direct oversight of mammography facilities. Every facility performing screening or diagnostic mammography must be accredited by an FDA-approved accreditation body (the American College of Radiology is the largest; some states also accredit), hold an FDA certificate, and undergo annual inspections by FDA or state inspectors. MQSA regulates facilities and their use of equipment, while 21 CFR 1020 regulates what manufacturers build.

Medical Physicist Qualifications (21 CFR 900.12(a)(3))

• Initial qualifications:
  1. State licensed or approved, or certified in an appropriate specialty by a body FDA recognizes (such as the ABR or ABMP).
  2. A master's degree or higher in a physical science with at least 20 semester hours of physics.
  3. 20 contact hours of documented specialized training in conducting mammography surveys.
  4. Experience surveying at least 1 mammography facility and 10 mammography units, under the direct supervision of a qualified physicist (no more than one survey of a given unit in 60 days counts).
• Continuing education: at least 15 continuing education hours in mammography in the preceding 36 months.
• Continuing experience: surveys of at least 2 facilities and 6 units in the preceding 24 months.
• A physicist who falls short must regain the missing education or experience before surveying independently again.

---

The Annual Physicist Survey

Under 21 CFR 900.12(e)(9), each facility must have a survey by a medical physicist (or someone under the physicist's direct supervision) at least once a year. The survey covers the annual tests and the weekly phantom image test, the physicist evaluates the facility's own QC records and corrective actions, and the signed report must be sent to the facility within 30 days. A mammography equipment evaluation is also required whenever a unit or processor is installed, moved, or has major components changed (900.12(e)(10)).

The specific annual tests in 900.12(e)(5) were written for screen-film systems. For digital mammography and tomosynthesis, 900.12(e)(6) requires a QA program substantially the same as the image receptor manufacturer's (or the ACR Digital Mammography QC Manual where accepted), except that the dose may never exceed the screen-film limit. The screen-film tests remain the conceptual core:

1. Collimation and Alignment (900.12(e)(5)(vii))

• The x-ray field must reach the chest-wall edge of the image receptor and may not extend beyond any receptor edge by more than 2% of the SID.
• The light field and x-ray field may be misaligned by no more than 2% of the SID (total along length or width).
• The chest-wall edge of the compression paddle may not extend beyond the chest-wall edge of the receptor by more than 1% of the SID.

2. Beam Quality (900.12(e)(5)(iv))

The HVL must meet the minimum of 21 CFR 1020.30(m)(1), extended to the mammographic range in Table 2, which equals kVp/100 mm Al (for example, $0.30\text{ mm Al}$ at $30\text{ kV}$). The ACR's screen-film QC manual measures the HVL with the compression paddle in the beam and adds tighter guidance: a minimum of $\text{kVp}/100 + 0.03\text{ mm Al}$ and a maximum of $\text{kVp}/100 + C$, with $C = 0.12$ (Mo/Mo), $0.19$ (Mo/Rh), $0.22$ (Rh/Rh), and $0.30$ (W/Rh). An HVL that is too high signals excess filtration and lost calcification contrast.

3. kVp Accuracy and Reproducibility (900.12(e)(5)(ii))

• kVp must be accurate within ±5% at the lowest measurable, most commonly used, and highest clinical settings.
• At the most common setting, the coefficient of variation must be ≤ 0.02.

4. AEC Performance and Reproducibility (900.12(e)(5)(i) and (v))

• For screen-film, the AEC must hold optical density within ±0.15 of the mean as homogeneous phantom thickness varies from 2 to 6 cm, with a phantom-center density of at least 1.20.
• The coefficient of variation of both air kerma and mAs must not exceed 0.05.

5. Compression (900.12(e)(4)(iii), semiannual)

The compression device must provide at least 111 N (25 lb), and the maximum force of the initial power drive must be between 111 N (25 lb) and 200 N (45 lb).

6. System Resolution (screen-film)

With a line-pair pattern 4.5 cm above the breast support, at least 11 lp/mm with bars perpendicular to the anode-cathode axis and 13 lp/mm with bars parallel to it.

---

Mean Glandular Dose (MGD)

Biological Basis

The tissue at risk for radiation-induced breast cancer is the glandular (fibroglandular) tissue, so entrance skin dose is a poor risk indicator. Mammography uses the mean (average) glandular dose.

The MQSA Dose Limit (900.12(e)(5)(vi))

The average glandular dose delivered during a single craniocaudal view of an FDA-accepted phantom simulating a standard breast (about $4.2\text{ cm}$ compressed, roughly 50% glandular) must not exceed:

$\mathbf{3.0\text{ mGy } (0.3\text{ rad}) \text{ per exposure}}$

using the technique factors and conditions used clinically for a standard breast. Typical digital units deliver about $1$ to $2\text{ mGy}$ per view.

Calculating MGD

In the Dance formalism, used by European and IAEA protocols and many US physicists:

$D_g = K \cdot g \cdot c \cdot s$

• $K$ is the incident air kerma at the upper surface of the breast (or phantom), measured free in air without backscatter.
• $g$ converts incident air kerma to MGD for a 50% glandular breast; it depends on breast thickness and HVL.
• $c$ corrects for glandularity other than 50%.
• $s$ corrects for target/filter combinations other than Mo/Mo: about $1.000$ (Mo/Mo), $1.017$ (Mo/Rh), $1.061$ (Rh/Rh), and $1.042$ (W/Rh) (Dance et al., 2000); values for W/Ag and W/Al are tabulated separately.

US accreditation manuals historically used equivalent Wu/Sobol "DgN" tables, which convert entrance exposure to MGD in one step.

---

ACR Phantom Scoring

Legacy ACR Mammography Phantom

The legacy phantom (simulating a 4.2 cm, 50/50 breast) contains a wax insert with:

• 6 nylon fibers: $1.56, 1.12, 0.89, 0.75, 0.54, 0.40\text{ mm}$ diameter.
• 5 speck groups (aluminum oxide): $0.54, 0.40, 0.32, 0.24, 0.16\text{ mm}$, six specks per group.
• 5 masses: $2.00, 1.00, 0.75, 0.50, 0.25\text{ mm}$ thick.

Objects are scored from the largest down: a fiber scores 1 when its full length is seen in the correct location and orientation and 0.5 when more than half is seen; a speck group scores 1 for 4–6 specks and 0.5 for 2–3; a mass scores 1 for a density with a circular border and 0.5 for a density without one. Counting stops at the first object scoring 0 or 0.5 (a 0.5 is added before stopping), and artifacts that look like objects are deducted. The passing score is 4 fibers, 3 speck groups, and 3 masses.

ACR Digital Mammography Phantom

The ACR Digital Mammography QC Manual (2018) introduced a larger phantom with 6 fibers, 6 speck groups, and 6 masses of smaller sizes. It passes when the 2 largest fibers, 3 largest speck groups, and 2 largest masses are seen with no clinically significant artifacts. The two passing scores are not interchangeable because the objects differ.

---

Digital Breast Tomosynthesis (DBT)

In DBT the tube sweeps through a limited arc (about $15^\circ$ to $50^\circ$, depending on the vendor), taking roughly 9 to 25 low-dose projections that are reconstructed into thin slices, usually 1 mm apart. Physicist testing follows the manufacturer's or ACR's DBT QC program and includes z-axis resolution or artifact spread, volume coverage, AEC, collimation, and dose.

• Combination mode: acquiring both a full-field digital mammogram and a DBT sweep for each view roughly doubles the dose compared with either alone.
• Synthetic 2D: a 2D image computed from the DBT data replaces the separate 2D exposure, bringing the total dose back to that of the DBT acquisition alone.

---

Worked Numerical Examples

Example 1: Mean Glandular Dose

A W/Rh digital unit images the accreditation phantom at $28\text{ kVp}$ with an HVL of $0.54\text{ mm Al}$ and an incident air kerma of $K = 3.90\text{ mGy}$. For a 4.2 cm, 50% glandular breast at this HVL, the Dance tables give $g \approx 0.27$; $s = 1.042$ for W/Rh and $c = 1.00$.

$D_g = 3.90 \times 0.27 \times 1.00 \times 1.042 = 1.10\text{ mGy}$

This is well below the 3.0 mGy MQSA limit.

Example 2: Beam Quality Check

A Mo/Mo unit at $30\text{ kVp}$ has an HVL of $0.35\text{ mm Al}$ measured through the compression paddle.

• MQSA minimum (Table 2): $30/100 = 0.30\text{ mm Al}$, so it passes.
• ACR screen-film manual minimum: $0.30 + 0.03 = 0.33\text{ mm Al}$, so it passes.
• ACR maximum: $0.30 + 0.12 = 0.42\text{ mm Al}$, so it passes.

---

Pitfalls

• Leaving the paddle out of HVL measurements: the ACR procedure measures through the paddle, and omitting it lowers the HVL and can make the unit appear to fail.
• Applying screen-film criteria to digital units: digital systems follow the manufacturer's (or ACR digital) QC program, apart from the dose limit.
• Ignoring the spectral factor: using $s = 1.000$ for a tungsten-target system underestimates MGD by several percent.
• Mixing phantom pass criteria: 4-3-3 applies to the legacy phantom and 2-3-2 to the ACR Digital Mammography phantom.
• Assuming DBT always doubles dose: synthetic 2D avoids the separate 2D exposure.