Flat-Panel Detector QC: Uniformity & Dead Pixels

Flat-panel detector quality control confirms that a digital radiography receptor produces a uniform, low-noise image with an acceptable number of defective pixels — so that detector artifacts are never mistaken for, or allowed to hide, real anatomy. The core receptor-level tests are signal nonuniformity, signal-to-noise ratio (SNR) nonuniformity, anomalous or defective pixel counts, spatial resolution, and checks for lag and ghosting, evaluated with the standardized methods of AAPM Task Group 150 and TG-151. ¹²³

Unlike film or the exposure-related metrics that dominate dose-optimization discussions, detector QC asks a narrower question: is the receptor itself healthy? A digital detector can produce a technically "correct" exposure while quietly developing dead columns, a stale gain calibration, or residual ghosting that degrades every image it acquires. This guide explains what those failure modes are, how physicists measure them, what tolerances and frequencies are appropriate, and how DRPS integrates detector QC into a defensible diagnostic radiography physics program.

Introduction

Modern general radiography is almost entirely digital, built on either indirect-conversion flat-panel detectors (a scintillator such as cesium iodide coupled to a photodiode array), direct-conversion detectors (a photoconductor such as amorphous selenium), or computed radiography (CR) storage-phosphor plates. In every case, the image is formed from millions of individual detector elements (dexels), each of which must respond predictably to X-rays and be corrected by an onboard calibration before a diagnostic image is displayed. ¹⁴

That calibration is both the strength and the weakness of digital detectors. Flat-field (gain) and offset (dark) corrections can make a physically imperfect detector look excellent — until the calibration drifts, a region of the array fails, or the correction itself bakes in an artifact. Because the display software stretches contrast automatically, subtle receptor problems are not always obvious to the technologist or radiologist at the workstation. QC exists to surface those problems on purpose, with quantitative flat-field images, before they reach a patient's study.

AAPM Task Group 150 provides the physicist-level acceptance and performance framework for digital radiographic systems, including standardized, region-of-interest (ROI)-based measurements of nonuniformity, minimum SNR, anomalous pixels, and modulation transfer function (MTF). ¹³ AAPM Task Group 151 defines the ongoing, higher-frequency QC program — rejected-image analysis, exposure analysis, and artifact identification — carried out by technologists under a physicist's direction. ² Together with IEC detector-characterization and constancy standards, they form the backbone of a defensible detector QC program. ⁴⁵⁷

Topic Explanation

What is flat-panel detector quality control?

Flat-panel detector QC is the subset of digital radiography testing that evaluates the imaging receptor's intrinsic performance rather than the X-ray tube or generator. It confirms that, after the detector applies its offset and gain corrections, the resulting image is spatially uniform, adequately low in noise, free of unacceptable defective pixels, spatially sharp, and free of time-dependent residual signal.

The receptor-level questions a QC program must answer include:

• Is the signal uniform across the active area when the detector is exposed to a flat, uniform beam?
• Is the noise (and therefore SNR) uniform, or are there regions that are noisier than others?
• How many anomalous or defective pixels exist, are they clustered, and are they increasing over time?
• Is spatial resolution (MTF) stable and within specification?
• Does residual signal (lag) and gain change (ghosting) from a prior exposure decay adequately before the next image?

These are distinct from — though complementary to — the exposure-side metrics covered in our guide to the digital radiography exposure index and the frequency-domain performance covered in detective quantum efficiency. Exposure index tells you whether the technique factors were appropriate; DQE tells you how efficiently the detector converts dose into image information; uniformity and defective-pixel QC tell you whether the receptor hardware and its calibration are actually healthy today.

Key detector failure modes

A digital detector can degrade in several characteristic ways, and QC is designed to separate them:

• Stale or failed gain calibration. Flat-field calibration corrects for pixel-to-pixel gain differences, scintillator thickness variation, and the anode heel effect present at calibration. As the tube ages or the calibration expires, residual structured nonuniformity appears.
• Dead, stuck, or drifting pixels. Individual dexels can stop responding, saturate, or wander. Isolated defects are interpolated automatically; clusters and lines are not always handled gracefully.
• Defective rows/columns and readout electronics. Failures in the thin-film-transistor readout or data lines produce line artifacts that can run the full length of the panel.
• Lag and ghosting. Charge trapping leaves residual signal (lag) in subsequent frames; prior irradiation can change local gain (ghosting), altering the detector's later response. ¹
• Physical and environmental damage. Portable and wireless detectors are dropped, flexed, and exposed to fluids; moisture ingress and mechanical damage produce localized artifacts.

The clinical danger is not the existence of these defects — every detector has some — but their being invisible at the workstation while subtly mimicking a nodule, fracture line, or line/tube, or masking one. QC makes them visible and trends them so the physicist can distinguish "recalibrate the detector" from "the panel is failing and must be replaced."

Key Technical Principles

The receptor test set at a glance

The table below summarizes the principal receptor-level QC tests, what each detects, how it is typically acquired, and a representative frequency. Specific tolerances are set by the manufacturer specification, AAPM TG-150/TG-151 methodology, and the facility's accreditation program; the values here describe the type of criterion, not a universal pass/fail number. ¹²³

Signal nonuniformity

For a uniform flat-field exposure, the image is divided into a grid of ROIs and the mean signal ${\bar{S}}_i$ in each ROI is compared. A widely used global signal nonuniformity metric expresses the spread of ROI means relative to the overall mean:

$$U_{\text{signal}}=\frac{S_{\max }-S_{\min }}{S_{\text{mean}}}\times 100\%$$

where $S_{\max }$ and $S_{\min }$ are the largest and smallest ROI mean values and $S_{\text{mean}}$ is the average across all ROIs. ³ A local nonuniformity version compares each ROI to its immediate neighbors, which is more sensitive to small, sharply bounded defects than the global metric.

Worked example (illustrative values): suppose a flat-field ROI grid yields a maximum ROI mean of 2,520, a minimum of 2,410, and an overall mean of 2,480 (arbitrary detector units). Then:

$$U_{\text{signal}}=\frac{2520-2410}{2480}\times 100\%\approx 4.4\%$$

Whether 4.4% is acceptable depends on the ROI size, the manufacturer specification, and the facility's baseline — which is precisely why standardized ROI geometry and a stable baseline matter. Published TG-150 implementation work has shown that nonuniformity and minimum-SNR results depend on ROI size, so a fixed ROI layout should be used for consistency, and where the detector cannot be tested at its exact calibration position a baseline should be established from the mean of several repeated measurements. ³

SNR and SNR nonuniformity

Within each ROI, the signal-to-noise ratio is the mean divided by the standard deviation:

$${\text{SNR}}_i=\frac{{\bar{S}}_i}{{\sigma }_i}$$

Analogous to signal nonuniformity, SNR nonuniformity captures how much the SNR varies across the field:

$$U_{\text{SNR}}=\frac{{\text{SNR}}_{\max }-{\text{SNR}}_{\min }}{{\text{SNR}}_{\text{mean}}}\times 100\%$$

SNR nonuniformity is a sensitive detector-health indicator because a region can retain an acceptable mean signal while becoming noisier — an early sign of scintillator or photodiode degradation that pure signal-uniformity analysis can miss. Independent implementations of the TG-150 metrics have reported that the SNR-nonuniformity metric is more sensitive to detector nonuniformity than some vendor QC tools, underscoring its diagnostic value. ³

Anomalous and defective pixels

A defective (anomalous) pixel is one whose value deviates from its local expectation by more than a defined threshold in a flat-field or offset image. TG-150 methodology counts anomalous pixels using ROI-based comparisons and reports both isolated defects and clusters; importantly, the same work found that ROI size affects signal and SNR nonuniformity but not the detection of anomalous pixels, so defective-pixel counting is relatively robust to analysis geometry. ³

The clinically important quantities are:

• the total number of defective pixels versus the manufacturer's allowed maximum;
• the presence of clusters (adjacent defects) and line/column defects, which correction algorithms handle poorly;
• the trend over time — a slowly growing defect map is a leading indicator of panel failure.

Lag and ghosting

Lag is residual signal from a prior exposure appearing in a later frame; ghosting is a change in local gain induced by prior irradiation. Both are evaluated by acquiring a sequence of exposures and quantifying the residual. A simple residual-signal ratio compares the residual in a "dark" frame acquired shortly after a bright exposure to that bright signal:

$$R_{\text{lag}}=\frac{S_{\text{residual}}}{S_{\text{bright}}}\times 100\%$$

Acceptable residual-signal behavior is specified by the manufacturer and verified at acceptance and after detector service. ¹ In routine clinical use, adequate inter-exposure timing and the detector's own correction manage lag; QC confirms the residual decays as designed rather than accumulating into visible ghost images.

Clinical Impact

A healthy-looking exposure index does not guarantee a healthy detector. Because digital display windows contrast automatically, a detector with a stale gain map, a developing column defect, or regional SNR loss can still produce images that pass casual inspection while degrading diagnostic performance for every patient scanned on that unit.

Concrete clinical consequences include:

• False positives. A cluster of defective pixels or a calibration artifact can resemble a pulmonary nodule, a subtle fracture, or free air, prompting unnecessary follow-up imaging and dose.
• False negatives. Regional noise elevation or shading can mask low-contrast findings, especially in high-attenuation regions where SNR is already marginal.
• Line-and-tube confusion. Column defects and residual ghosts can be mistaken for catheters, leads, or hardware in portable chest and abdominal radiographs — exactly the setting where portable/wireless detectors are most abused. See our companion guidance on mobile radiography radiation safety for the handling realities that drive detector damage.
• Repeat exposures. Artifacts that technologists notice but cannot resolve drive repeat images, which is why detector QC and repeat/reject analysis belong in the same program.

Because the failures are gradual and the display hides them, the value of detector QC is early, quantitative detection: catching the drift before it becomes a diagnostic error, and distinguishing a five-minute recalibration from a costly panel replacement.

Practical Optimization Tips

Establish and defend a baseline

Every quantitative detector metric is only meaningful relative to a baseline captured at acceptance, when the detector is known-good. Record the flat-field technique (kVp, added filtration, air kerma at the detector), the ROI geometry, and the resulting nonuniformity, SNR, anomalous-pixel count, and MTF. Because nonuniformity metrics depend on ROI size and exposure geometry, freeze those parameters and reuse them exactly at every subsequent evaluation. ³

Test the way the detector is used

• Use a reproducible flat-field setup — consistent beam quality, adequate and uniform field coverage, and, where possible, the manufacturer's calibration geometry.
• For wireless and portable detectors, test them in the state they are used, including any grid and the actual acquisition workstation, not on a bench in an ideal position.
• Acquire an offset/dark image as well as a bright flat-field so offset drift and fixed-pattern noise are captured, not just gain-related shading.

Recalibrate before you condemn

Many "nonuniformity failures" are simply an expired or heel-effect-contaminated gain calibration. A structured protocol is: (1) review the artifact, (2) run offset and gain calibration per the manufacturer, (3) re-acquire the flat-field, and (4) only then decide whether a residual defect is true hardware failure. Document the before/after so the decision is defensible.

Common pitfalls to avoid

• Changing ROI size or technique between visits, which makes trends meaningless.
• Averaging away a local defect by relying only on the global nonuniformity number — always inspect the flat-field image and use a local metric.
• Ignoring the defective-pixel trend because "it still passes" — a growing map predicts failure.
• Treating vendor QC tools as interchangeable. Manufacturer utilities (for example, vendor quality-assurance processes and total-quality tools) use different noise and beam-quality assumptions than TG-150, so their numbers are not directly comparable and should be trended against themselves, with the physicist's standardized test as the independent check. ³
• Confusing exposure problems with detector problems. An abnormal exposure index is a technique/AEC issue, addressed in our automatic exposure control guide, not necessarily a receptor failure.

Regulatory Considerations

Detector QC lives inside a broader regulatory and accreditation framework for radiographic equipment, even though no single federal rule prescribes a universal detector-uniformity tolerance. The applicable structure combines federal equipment standards, state radiation-control rules, national physics standards, and accreditation requirements.

• FDA 21 CFR Part 1020. Diagnostic X-ray systems and their major components are subject to federal performance standards under 21 CFR 1020.30 and 1020.31; the X-ray machine itself is regulated by the FDA and by state radiation-control programs rather than by the NRC (which governs radioactive material). ⁹
• State radiation-control programs. States require registration of radiation-producing machines and, in many cases, periodic medical-physicist evaluation. In the states DRPS serves — Florida, Maryland, Virginia, Washington DC, California, Nevada, Pennsylvania, New York, New Jersey, and Delaware — the machine-source requirements are administered by the state program; always confirm the current rule with the authority having jurisdiction.
• ACR–AAPM Technical Standard. The ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of Radiographic Equipment specifies that radiographic equipment performance be evaluated upon installation and monitored at least annually by a qualified medical physicist, with a continuous QC program in place; its appendix enumerates the parameters — including image-receptor performance — to be assessed. ⁸
• AAPM and IEC methodology. AAPM Report No. 150 (TG-150) and TG-151 define the physicist and technologist test programs, respectively, while IEC 62220-1-1:2015 (DQE), IEC 62494-1:2008 (exposure index), and IEC 61223-3-8:2024 (acceptance and constancy tests for radiographic/radioscopic imaging performance) provide the internationally standardized measurement basis. ¹²⁴⁵⁷
• Accreditation. Facilities pursuing or maintaining ACR accreditation must demonstrate an ongoing QC program and satisfactory physics testing; detector performance is part of that evidence. Our overview of ACR accreditation physics requirements puts detector QC in the accreditation context. ¹⁰

The practical takeaway: detector QC results should be documented, trended, and tied to the annual physics report so they are defensible during accreditation review and state inspection.

Frequently Asked Questions (FAQs)

What is flat-panel detector quality control?

Flat-panel detector quality control is the set of tests that confirm a digital radiography receptor produces a uniform, low-noise image with an acceptable number of defective pixels. It typically includes signal nonuniformity, SNR nonuniformity, anomalous or defective pixel counts, spatial resolution (MTF), and checks for lag and ghosting, following AAPM TG-150 and TG-151 methodology.

How often should flat-panel detector uniformity be tested?

Physicist-level detector performance tests such as nonuniformity, SNR nonuniformity, anomalous pixels, and MTF are commonly evaluated at acceptance and at least annually, while technologist-level constancy checks such as uniformity review, exposure-index trending, and artifact inspection are performed more frequently — often daily to monthly — under the direction of a qualified medical physicist.

What causes flat-panel detector nonuniformity?

Nonuniformity can result from an outdated or failed gain (flat-field) calibration, dead or drifting pixels and readout columns, scintillator or photodiode defects, charge trapping and lag, heel-effect residue in the calibration, moisture ingress, or physical damage from dropping a portable detector. QC distinguishes correctable calibration problems from true hardware failure.

Are defective pixels in a digital detector a safety problem?

A small number of isolated defective pixels is expected and is corrected by interpolation, so it is usually not clinically significant. The concern is clustered defects, defective lines or columns, and defects that grow over time, because these can mimic or mask pathology. QC tracks the defective-pixel map against the manufacturer specification and trends it over time.

What is the difference between detector lag and ghosting?

Lag is residual signal carried over from a previous exposure that appears in the next image, while ghosting is a change in detector gain caused by prior irradiation that alters the response to later exposures. Both are time-dependent detector effects, and QC checks that residual signal decays adequately between exposures.

Does flat-panel detector QC replace the annual physics survey?

No. Detector QC is one part of a complete radiographic physics evaluation that also covers tube output, kVp accuracy, half-value layer, automatic exposure control, collimation, and dose. Detector tests confirm the imaging receptor is performing, but the full system must be evaluated together against ACR–AAPM and state requirements.

Can a facility perform detector QC in-house?

Technologists can perform constancy and artifact checks under a medical physicist's direction using manufacturer QC tools, but acceptance testing and the annual quantitative detector evaluation should be performed or overseen by a qualified or board-certified medical physicist so results are defensible for accreditation and inspection.

Key Takeaways

• Detector QC asks whether the receptor is healthy, separate from whether the exposure was correct — a good exposure index does not prove a good detector.
• The core receptor tests are signal nonuniformity, SNR nonuniformity, anomalous pixels, MTF, and lag/ghosting, standardized by AAPM TG-150 (physicist) and TG-151 (ongoing/technologist). ¹²³
• Standardize ROI geometry and technique and freeze a baseline, because nonuniformity and SNR metrics depend on ROI size and exposure geometry. ³
• Trend the defective-pixel map; clusters, lines, and growth predict panel failure even when the count still "passes."
• Recalibrate before condemning — many nonuniformity failures are a stale gain map, not dead hardware.
• Tie results to the ACR–AAPM annual evaluation and accreditation evidence, and confirm state machine-source requirements with the authority having jurisdiction. ⁸⁹¹⁰

Conclusion

Flat-panel detectors are engineered to hide their own imperfections behind offset and gain corrections and automatic display windowing. That engineering is what makes digital radiography robust — and what makes disciplined detector QC essential. Signal and SNR nonuniformity, anomalous-pixel mapping, MTF, and lag/ghosting checks give the medical physicist a quantitative, trendable picture of receptor health that the workstation image alone will not reveal.

A strong program combines physicist-level acceptance and annual testing with technologist-level constancy checks, uses standardized TG-150 methodology and a frozen baseline, and documents results against ACR–AAPM and accreditation requirements. Done well, it catches drift early, prevents artifacts from being read as anatomy, and turns detector replacement into a planned decision rather than an emergency.

How DRPS Can Help

Diagnostic Radiation Physics Services performs acceptance testing and annual diagnostic radiography physics evaluations that include quantitative flat-panel detector QC — signal and SNR nonuniformity, defective-pixel mapping, MTF, and lag/ghosting — using standardized TG-150 methodology. We help facilities establish defensible baselines, set up technologist constancy programs under physicist direction, interpret vendor QC output, and align documentation with accreditation support and medical physicist consulting needs.

DRPS supports imaging facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware. To scope a detector QC or annual physics evaluation, contact our team.

Related Resources

• Digital radiography exposure index
• Detective quantum efficiency in digital radiography
• Repeat/reject analysis
• Automatic exposure control in radiography
• Mobile radiography radiation safety
• ACR accreditation physics requirements
• Diagnostic radiography physics testing
• Accreditation support

References

1. American Association of Physicists in Medicine. Acceptance Testing and Quality Control of Digital Radiographic Imaging Systems. AAPM Report No. 150 (Task Group 150). College Park, MD: AAPM. aapm.org
2. Jones AK, Heintz P, Geiser W, et al. Ongoing quality control in digital radiography: Report of AAPM Imaging Physics Committee Task Group 151. Medical Physics. 2015;42(11):6658-6670. doi:10.1118/1.4932623. PubMed
3. Li G, Greene TC, Nishino TK, Willis CE. Evaluation of cassette-based digital radiography detectors using standardized image quality metrics: AAPM TG-150 Draft Image Detector Tests. Journal of Applied Clinical Medical Physics. 2016;17(5):391-417. doi:10.1120/jacmp.v17i5.6008. PubMed
4. International Electrotechnical Commission. IEC 62220-1-1:2015 — Medical electrical equipment: Characteristics of digital X-ray imaging devices — Part 1-1: Determination of the detective quantum efficiency — Detectors used in radiographic imaging. Geneva: IEC; 2015. iec.ch
5. International Electrotechnical Commission. IEC 62494-1:2008 — Medical electrical equipment: Exposure index of digital X-ray imaging systems — Part 1: Definitions and requirements for general radiography. Geneva: IEC; 2008. iec.ch
6. Shepard SJ, Wang J, Flynn M, et al. An exposure indicator for digital radiography: AAPM Task Group 116 (executive summary). Medical Physics. 2009;36(7):2898-2914. PubMed
7. International Electrotechnical Commission. IEC 61223-3-8:2024 — Evaluation and routine testing in medical imaging departments — Part 3-8: Acceptance and constancy tests — Imaging performance of X-ray equipment for radiography and radioscopy. Geneva: IEC; 2024. iec.ch
8. American College of Radiology; American Association of Physicists in Medicine. ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of Radiographic Equipment. Reston, VA: ACR. acr.org
9. U.S. Food and Drug Administration. 21 CFR Part 1020 — Performance Standards for Ionizing Radiation Emitting Products (§§ 1020.30–1020.31, diagnostic X-ray systems and radiographic equipment). ecfr.gov
10. American College of Radiology. Digital Radiography (DR) / Computed Radiography (CR) Accreditation Program requirements. acraccreditation.org