What is the presampling MTF and why is it reported?

The presampling MTF characterizes the blurring of the imaging chain before the detector's discrete sampling is applied. It is measured with a slightly angled edge or slit so the edge is oversampled, which lets the physicist characterize the system beyond the pixel-limited Nyquist frequency and separate true blur from sampling and aliasing effects.

How does detector pixel pitch set the resolution limit?

The sampling pixel pitch sets the Nyquist frequency at one over twice the pitch. A 100-micrometer detector element gives a Nyquist frequency of 5 cycles per millimeter, 70 micrometers gives about 7.1, and 50 micrometers gives 10. Spatial-frequency content above Nyquist cannot be represented faithfully and can alias into lower frequencies.

Mammography MTF & Spatial Resolution QC

Q: Which test device is preferred for measuring mammography MTF?

Both slit and edge devices are established. An angled, attenuating edge is widely used because it tolerates small alignment errors far better than a narrow slit and needs less exposure. The edge spread function is differentiated to a line spread function, which is Fourier transformed to the MTF.

Q: Does the ACR Digital Mammography QC Manual require a technologist MTF test?

The routine ACR Digital Mammography QC program emphasizes phantom image quality, signal-to-noise and contrast-to-noise ratio, uniformity, and artifact evaluation rather than a daily MTF measurement. Detailed MTF and spatial-resolution characterization is typically part of the medical physicist's acceptance and annual survey work, and of manufacturer performance specifications.

Mammography spatial resolution is the system's ability to render fine, high-contrast detail — the edges of microcalcifications and the spicules of a small mass — and the modulation transfer function (MTF) is its rigorous, frequency-by-frequency description. A single "limiting resolution" number in line pairs per millimeter is convenient shorthand, but it hides where a detector actually loses detail. The MTF shows the full curve, and in digital mammography it must be read together with the detector's sampling pitch, which fixes the Nyquist frequency and governs aliasing.¹²³

This guide explains what the MTF is, how the presampling MTF is measured with an edge test device, how pixel pitch and the Nyquist limit constrain what a full-field digital mammography (FFDM) detector can resolve, and how all of this connects to the acceptance and annual survey work a medical physicist performs under the ACR Digital Mammography Quality Control program and the Mammography Quality Standards Act (MQSA).⁴⁵⁶

Introduction

In mammography, small-object detection depends on preserving high-spatial-frequency contrast, and the MTF is the metric that tells you whether the system does so. Breast cancer is frequently detected through clustered microcalcifications a few hundred micrometers across and through subtle margin detail. Those features live at high spatial frequencies, precisely where every imaging chain — focal spot, geometry, scintillator or photoconductor, and sampling — loses signal fastest.¹³

Screen-film mammography was historically specified by a limiting spatial resolution read from a bar pattern, often in the range of a dozen or more line pairs per millimeter. Digital mammography changed the question. A discretely sampled detector has a hard sampling ceiling set by its element spacing, and resolution is no longer a single bar-pattern number but a curve — the MTF — that must be interpreted alongside the detector's Nyquist frequency, its noise power spectrum, and its detective quantum efficiency (DQE).¹²³

This article focuses on the resolution half of that picture: what the MTF means, how it is measured defensibly, what pixel pitch and aliasing do to it, and how a physicist uses it during acceptance testing and the annual MQSA survey. For the noise-and-dose side of digital mammography QC, see our companion guides to mammography CNR and SDNR QC and the ACR digital mammography phantom QC.

Topic Explanation

What is the MTF?

The MTF describes the fraction of an object's contrast (modulation) that survives at each spatial frequency as it passes through the imaging system. Imagine imaging sinusoidal patterns of increasing frequency, all at the same input contrast. At low frequencies — coarse structures — the system reproduces nearly all of the contrast, so the MTF is close to 1. As the pattern gets finer, blurring reduces the output contrast, and the MTF falls toward 0.

Formally, modulation at a given spatial frequency is defined from the maximum and minimum signal:

M = \frac{S _{m a x} - S _{m i n}}{S _{m a x} + S _{m i n}}

and the MTF at frequency $f$ is the output modulation divided by the input modulation, normalized to unity at zero frequency:

MTF (f) = \frac{M _{out} ( f )}{M _{in} ( f )}

Because the imaging chain is, to good approximation, a linear and shift-invariant blurring operation, the MTF is also the magnitude of the Fourier transform of the system's line spread function (LSF), the one-dimensional response to an infinitely thin line input.¹⁷

Key resolution terms

Line spread function (LSF) — the system's blurred response to a thin line; its width quantifies blur.
Edge spread function (ESF) — the response to a sharp edge; its derivative is the LSF.
Presampling MTF — the MTF of the imaging chain before the detector's discrete sampling is imposed, obtained by oversampling an angled edge so the result is not limited by a single pixel row.¹⁷
Nyquist frequency ( $f_{N y}$ ) — the highest spatial frequency a detector of a given sampling pitch can represent without aliasing.
Aliasing — the misrepresentation of frequency content above Nyquist as spurious lower-frequency signal.

Why digital mammography is sampling-limited

A digital detector records signal only at discrete element locations separated by the sampling pitch $a$ . Sampling theory sets the highest faithfully recorded frequency — the Nyquist frequency — at:

f_{N y} = \frac{1}{2 a}

For the pixel pitches used in FFDM, this ceiling lands squarely in the range where clinically important detail lives, which is why detector element size is one of the defining specifications of a mammography system.²³ For background on exposure-side detector behavior, see our guide to the digital radiography exposure index.

Key Technical Principles

From edge image to MTF

The most common defensible route to the presampling MTF uses an attenuating edge placed near the detector and rotated a few degrees relative to the pixel matrix. The slight angle means successive pixel rows sample the edge at slightly different sub-pixel positions, so combining many rows synthesizes a finely oversampled edge spread function. The processing chain is:⁷⁸

Acquire the angled-edge image and determine the precise edge angle.
Reproject pixel values onto an axis perpendicular to the edge, producing a finely sampled, oversampled ESF.
Differentiate the ESF to obtain the LSF:

LSF (x) = \frac{d}{d x} ESF (x)

Fourier transform the LSF and take the normalized magnitude to obtain the presampling MTF:

MTF (f) = \frac{\int LSF ( x ) e ^{- 2 π i f x} d x}{\int LSF ( x ) d x}

An angled slit can be used in place of the edge, but the slit method demands precise fabrication and alignment and higher exposure, whereas the edge method tolerates misalignment of several centimeters along the edge and still returns an accurate curve.⁷⁸ An intercomparison of slit, translucent-edge, and opaque-edge techniques found that the measurement method itself changes the estimated MTF by a few percent, so the physicist should keep the device, beam quality, and processing fixed across surveys to make the comparison to baseline meaningful.⁸

Worked example: Nyquist and the resolution ceiling

Consider three FFDM detectors with sampling pitches of 100, 70, and 50 micrometers. The Nyquist frequency follows directly from $f_{N y} = 1/ (2 a)$ :

f_{N y} (100 μ m) = \frac{1}{2 \times 0.100 mm} = 5.0 cycles/mm

f_{N y} (70 μ m) = \frac{1}{2 \times 0.070 mm} \approx 7.1 cycles/mm

f_{N y} (50 μ m) = \frac{1}{2 \times 0.050 mm} = 10.0 cycles/mm

These numbers are the ceilings, not the delivered resolution. The delivered resolution is the presampling MTF curve, which has already fallen well below 1 by the time it reaches Nyquist. For a clinical amorphous-silicon/cesium-iodide FFDM detector with 100-micrometer pixels, the measured presampling MTF was reported as approximately 0.73 at 2 cycles/mm, 0.42 at 4 cycles/mm, and 0.28 at 5 cycles/mm — meaning only about a quarter of the input contrast survives at the Nyquist frequency.³ A 50-micrometer detector pushes the ceiling to 10 cycles/mm, but whether that benefit is realized depends on the MTF at those frequencies and on whether the extra sampling simply records more noise.

Physics of the detector matters

Two detector technologies dominate FFDM, and their MTF behavior differs for physical reasons:¹⁹

Indirect-conversion (scintillator, e.g., cesium iodide on amorphous silicon). X-rays are first converted to light, which spreads laterally before detection. Structured (columnar) CsI limits this spread, but some light diffusion still reduces high-frequency MTF.
Direct-conversion (photoconductor, e.g., amorphous selenium). X-rays generate charge collected along field lines with little lateral spread, which tends to give a higher MTF at a given pitch. A subtlety specific to a-Se in the mammographic energy range is K-fluorescence reabsorption: because the selenium K-edge (about 12.7 keV) falls inside the mammographic spectrum, fluorescent photons can be reabsorbed a short distance away, and one analysis attributed roughly a 15% reduction in MTF at the Nyquist frequency to this effect.⁹

A comparison the physicist actually uses

Property	Powder CR	Indirect DR (CsI/a-Si)	Direct DR (a-Se)
Conversion path	X-ray → light (storage phosphor)	X-ray → light → charge	X-ray → charge
Typical sampling pitch	50 µm class	70–100 µm	70–85 µm
Dominant blur source	Light diffusion in phosphor + readout	Light spread in scintillator	Minimal lateral spread; K-fluorescence reabsorption
Reported presampling MTF at 5 cycles/mm	~0.08–0.23	higher of the DR group	up to ~0.64 across DR units
DQE at 5 cycles/mm (range across systems)	~0.02–0.20	—	up to ~0.41 across DR units

The MTF-at-5-cycles-per-millimeter and DQE-at-5-cycles-per-millimeter ranges above are drawn from a multi-system technical characterization of eleven digital mammography units (four CR and seven DR detectors), which found substantially higher high-frequency MTF and DQE for the DR detectors and the needle-phosphor CR system than for powder-phosphor CR.¹⁰ The table is a teaching comparison; the physicist's acceptance report should always use the specific unit's measured curves against its own baseline and the manufacturer's specification, not a generic range.

MTF, NPS, and DQE belong together

Resolution never travels alone. The MTF quantifies signal transfer; the noise power spectrum (NPS) quantifies how noise is distributed across spatial frequency; and the DQE combines both to describe how efficiently the detector uses the incident X-ray quanta:¹³¹⁰

DQE (f) = \frac{MTF ^{2} ( f )}{NNPS ( f ) q ˉ}

where $NNPS (f)$ is the normalized NPS and $\overset{q}{ˉ}$ is the incident photon fluence per unit exposure. A detector can have a respectable MTF but mediocre DQE if it is noisy, so acceptance testing characterizes all three. The IEC standards define the measurement conditions: IEC 62220-1 established the general DQE methodology, and IEC 62220-1-2 specifies the method for detectors used in mammography.¹¹¹²

Clinical Impact

Spatial-resolution performance translates directly into whether a reader can see and characterize the smallest clinically relevant findings. Microcalcifications, their morphology, and the fine margins of small masses are the features most sensitive to MTF degradation, and they are exactly the features that drive early detection and accurate BI-RADS characterization.¹³

A detector with a higher high-frequency MTF preserves microcalcification edge sharpness and the conspicuity of fine spiculation. But more sampling is not automatically better: pushing pixel pitch smaller raises the Nyquist ceiling and can improve high-frequency MTF, yet it also records more noise per pixel and can reduce DQE if the detector is dose-starved. The clinical sweet spot is the pitch and detector physics that maximize task-relevant information per unit glandular dose — which is why the physicist evaluates MTF, NPS, and DQE together rather than chasing resolution alone.³¹⁰

Resolution also degrades in ways that are invisible on a casual look at a clinical image but obvious on an MTF curve: detector element drift, scintillator damage, charge-trapping layers in a-Se detectors, or a reconstruction/processing change can all pull the curve down. Tracking the MTF against baseline is how a physicist catches gradual degradation before it reaches the clinical image.⁸⁹

Practical Optimization Tips

Measure the presampling MTF reproducibly

Use a well-characterized attenuating edge (or slit), rotated 1.5–3 degrees from the pixel matrix so the edge is oversampled.⁷⁸
Keep beam quality, added filtration, edge device, source-to-image distance, and the image-processing state identical across surveys. Method and beam-quality choices can shift the MTF by several percent, so consistency is what makes year-over-year comparison valid.⁸
Evaluate the MTF along both the chest-wall–to–nipple and left–right directions; detectors and reconstruction can be anisotropic.
Always report the detector's Nyquist frequency alongside the curve, and inspect behavior near and above Nyquist for aliasing.²³

Separate blur from sampling

Distinguish the presampling MTF (imaging-chain blur) from the pixel-limited sampling ceiling. A drop in the presampling MTF at mid frequencies points to detector or geometry problems; aliasing artifacts point to the sampling stage.¹⁷
Use the "for processing" (raw) image where the QC program and manufacturer specify it, because clinical "for presentation" processing alters frequency content and invalidates a physics MTF measurement.

Tie resolution QC to the whole program

Trend MTF, NPS/CNR, uniformity, and artifact findings together; a single metric rarely tells the whole story.¹⁰
Compare against the unit's acceptance baseline and the manufacturer's performance specification, not a generic number from the literature.
Fold geometry into the interpretation — focal-spot size and magnification affect system resolution upstream of the detector; see our guide to focal-spot size measurement.

Common pitfalls

Reporting a single "lp/mm" number and discarding the curve.
Measuring on processed (for-presentation) images.
Letting beam quality, filtration, or the edge device drift between surveys.
Ignoring the Nyquist frequency and mistaking aliasing for resolution.
Treating a higher sampling rate as proof of better imaging without checking DQE and dose.

Regulatory Considerations

Mammography is the most tightly regulated imaging modality in the United States, and spatial-resolution evaluation sits inside that framework rather than beside it. Under the Mammography Quality Standards Act, codified at 21 CFR Part 900, every facility performing mammography must be accredited and certified, must use equipment that meets quality standards, and must have a qualified medical physicist perform an annual equipment evaluation (the "annual survey") and acceptance testing of new units.⁴

The ACR Digital Mammography Quality Control Manual provides an FDA-accepted quality-control program for FFDM and digital breast tomosynthesis units.⁶ The medical physicist performs the manual's physicist-level tests — including image-quality, signal-to-noise, contrast-to-noise, and artifact evaluation — at least annually, while technologists perform the more frequent routine tests.⁵ The routine program emphasizes phantom image quality and quantitative signal metrics rather than a daily MTF measurement; detailed MTF and spatial-resolution characterization is generally handled at acceptance and as part of the physicist's survey and troubleshooting, and against manufacturer performance specifications. The ACR–AAPM technical standard for mammography describes the qualifications and scope of that physics work.¹³

Jurisdictionally, mammography units are radiation-producing X-ray machines regulated by the FDA under MQSA and by state radiation-control programs, not by the NRC. DRPS serves facilities across Florida, Maryland, Virginia, Washington DC, California, Nevada, Pennsylvania, New York, New Jersey, and Delaware; each state's radiation-control rules apply to the X-ray equipment in parallel with the federal MQSA requirements. Facilities should confirm both the federal MQSA obligations and the applicable state requirements with the authority having jurisdiction.

For the measurement conditions behind the physics numbers, the relevant international standards are IEC 62220-1 for the general DQE methodology and IEC 62220-1-2 for mammography detectors; these define beam qualities and procedures so that MTF, NPS, and DQE are reported on a common basis.¹¹¹²

Frequently Asked Questions (FAQs)

What is the modulation transfer function (MTF) in mammography?

The MTF is a frequency-resolved measure of how faithfully a mammography system transfers object contrast into the image at each spatial frequency. It describes the fraction of input modulation preserved as detail gets finer, so it is the quantitative description of spatial resolution rather than a single limiting number.¹

Why is the presampling MTF reported instead of a single resolution number?

The presampling MTF characterizes the blurring of the imaging chain before the detector's discrete sampling is applied. Measuring it with a slightly angled edge oversamples the response, which lets the physicist characterize the system beyond the pixel-limited Nyquist frequency and separate true blur from sampling and aliasing effects. A single bar-pattern number cannot do that.¹⁷

How does pixel pitch set the resolution limit?

The sampling pitch sets the Nyquist frequency at $f_{N y} = 1/ (2 a)$ . A 100-micrometer element gives 5 cycles/mm, 70 micrometers about 7.1, and 50 micrometers 10. Frequency content above Nyquist cannot be represented faithfully and can alias into lower frequencies.²³

Which test device is preferred for measuring mammography MTF?

Both slit and edge devices are established. An angled attenuating edge is widely used because it tolerates small alignment errors far better than a narrow slit and needs less exposure. The edge spread function is differentiated to a line spread function, which is Fourier transformed to the MTF.⁷⁸

Does the ACR Digital Mammography QC Manual require a technologist MTF test?

The routine program emphasizes phantom image quality, signal-to-noise and contrast-to-noise ratio, uniformity, and artifact evaluation rather than a daily MTF measurement. Detailed MTF and spatial-resolution characterization is typically part of the medical physicist's acceptance and annual survey work, and of manufacturer performance specifications.⁵

Who should perform mammography MTF and spatial-resolution testing?

A qualified or board-certified medical physicist performs acceptance and annual spatial-resolution evaluation as part of the mammography equipment survey required under MQSA and the ACR accreditation program, interpreting the MTF against baseline and manufacturer specifications.⁴⁵¹³

Key Takeaways

Spatial resolution in mammography is best described by the full MTF curve, not a single limiting-resolution number.¹
The presampling MTF is measured with an angled edge (or slit); the ESF is differentiated to an LSF and Fourier transformed to the MTF.⁷⁸
Detector pixel pitch sets the Nyquist frequency at $f_{N y} = 1/ (2 a)$ — 5, ~7.1, and 10 cycles/mm for 100-, 70-, and 50-micrometer pitches — and governs aliasing.²³
Detector physics matters: direct a-Se detectors generally show higher MTF than indirect CsI/a-Si at a given pitch, with a-Se K-fluorescence reabsorption reducing MTF near Nyquist by roughly 15%.⁹
MTF, NPS, and DQE must be interpreted together; more sampling is not automatically better imaging.³¹⁰
MTF and spatial-resolution characterization are part of the physicist's MQSA/ACR acceptance and annual survey work, measured reproducibly against baseline.⁴⁵¹³

How DRPS Can Help

Diagnostic Radiation Physics Services supports mammography facilities with acceptance testing of new FFDM and tomosynthesis units, annual MQSA equipment evaluations, MTF and spatial-resolution characterization, and troubleshooting of image-quality degradation — all performed and documented by board-certified medical physicists. Our mammography physics and MQSA survey services and broader medical physics consulting and accreditation support help facilities keep resolution performance defensible across the life of the equipment.

DRPS serves facilities across Florida, Maryland, Virginia, Washington DC, California, Nevada, Pennsylvania, New York, New Jersey, and Delaware.

Conclusion

The MTF turns "spatial resolution" from a single bar-pattern number into a physically meaningful curve that a physicist can measure, trend, and defend. In digital mammography, that curve has to be read alongside the detector's sampling pitch, its Nyquist frequency, and its noise and efficiency metrics. A reproducible presampling-MTF measurement with an edge device, interpreted against baseline and manufacturer specification and folded into the ACR and MQSA survey framework, is how a mammography program ensures that the fine detail driving early breast-cancer detection is actually being delivered to the reader.¹⁵¹⁰

Related Resources

References

Yaffe MJ, Rowlands JA. X-ray detectors for digital radiography. Physics in Medicine and Biology. 1997;42(1):1-39. doi:10.1088/0031-9155/42/1/001. iopscience.iop.org
Dobbins JT, Ergun DL, Rutz L, Hinshaw DA, Blume H, Clark DC. DQE(f) of four generations of computed radiography acquisition devices. Medical Physics. 1995;22(10):1581-1593. doi:10.1118/1.597627. PubMed
Vedantham S, Karellas A, Suryanarayanan S, et al. Full breast digital mammography with an amorphous silicon-based flat panel detector: physical characteristics of a clinical prototype. Medical Physics. 2000;27(3):558-567. doi:10.1118/1.598895. PubMed
U.S. Food and Drug Administration. Mammography Quality Standards Act (MQSA); 21 CFR Part 900. accessdata.fda.gov
American College of Radiology. 2018 ACR Digital Mammography Quality Control Manual. Reston, VA: ACR; 2018. acr.org
U.S. Food and Drug Administration. MQSA Alternative Standard #24: Approval of an Alternative Standard for Using the ACR Digital Mammography Quality Control Manual. fda.gov
Samei E, Flynn MJ, Reimann DA. A method for measuring the presampled MTF of digital radiographic systems using an edge test device. Medical Physics. 1998;25(1):102-113. doi:10.1118/1.598165. PubMed
Samei E, Ranger NT, Dobbins JT, Chen Y. Intercomparison of methods for image quality characterization. I. Modulation transfer function. Medical Physics. 2006;33(5):1454-1465. doi:10.1118/1.2188816. PubMed
Zhao W, Ji WG, Debrie A, Rowlands JA. Imaging performance of amorphous selenium based flat-panel detectors for digital mammography: characterization of a small area prototype detector. Medical Physics. 2003;30(2):254-263. doi:10.1118/1.1538233. PubMed
Marshall NW, Monnin P, Bosmans H, Bochud FO, Verdun FR. Image quality assessment in digital mammography: part I. Technical characterization of the systems. Physics in Medicine and Biology. 2011;56(14):4201-4220. doi:10.1088/0031-9155/56/14/002. PubMed
International Electrotechnical Commission. IEC 62220-1:2003, Medical electrical equipment — Characteristics of digital X-ray imaging devices — Part 1: Determination of the detective quantum efficiency. Geneva: IEC; 2003. webstore.ansi.org
International Electrotechnical Commission. IEC 62220-1-2:2007, Medical electrical equipment — Characteristics of digital X-ray imaging devices — Part 1-2: Determination of the detective quantum efficiency — Detectors used in mammography. Geneva: IEC; 2007. webstore.iec.ch
American College of Radiology. ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of Mammography Equipment. Reston, VA: ACR. acr.org