Digital Radiography Lag and Ghosting QC

Lag and ghosting are the two temporal artifacts of a flat-panel digital radiography detector: lag is residual signal carried over from a prior exposure, and ghosting is a change in detector sensitivity caused by exposure history. Both arise from charge trapping in the receptor, both can imitate or hide anatomy, and both belong in a defensible acceptance-testing and ongoing quality-control program. ¹²³

Unlike detector noise or spatial resolution, these are memory effects: the detector remembers the last thing it imaged. A bright bone edge, a pacemaker, or a lead marker from one exposure can leave a faint trace in the next patient's image if the receptor has not fully relaxed. Understanding why that happens — and how to measure it — is what separates a receptor that is genuinely artifact-free from one that merely looks fine on a single test image. ³⁴

Introduction

Digital radiography (DR) replaced film and computed radiography with flat-panel detectors that read out in seconds and are reused thousands of times per day. That reuse is the source of the problem. Every exposure deposits charge in the detector, and not all of that charge is cleared before the next exposure. The leftover charge produces two distinct phenomena that a qualified medical physicist must characterize and control. ³⁴

The first is lag — sometimes called image lag, temporal lag, or carry-over — which is genuine residual signal from a prior frame bleeding into later frames. The second is ghosting, a subtler effect in which prior exposure changes the detector's local gain, so a later uniform exposure is recorded non-uniformly even though no leftover signal is present. ¹³ The distinction is not academic: lag adds signal, ghosting changes sensitivity, and the two are measured and corrected in different ways. ¹

This guide walks through what lag and ghosting are, the charge-trapping physics that produces them in amorphous-silicon and amorphous-selenium receptors, how physicists quantify them, a worked measurement example, the clinical consequences, practical optimization tips, and how these tests fit into an AAPM TG-150 and TG-151 quality-control program. DRPS performs these evaluations as part of its diagnostic radiography physics testing and accreditation support services.

Topic Explanation

What are lag and ghosting?

Lag is residual signal, and ghosting is residual sensitivity change. A flat-panel detector converts X-ray energy into charge, stores that charge at each pixel, and reads it out. If the readout does not recover every stored electron, the leftover charge appears in the next frame as lag. If, instead, the prior exposure alters how efficiently the detector converts or collects charge in a region, that region will respond differently to the next exposure — that is ghosting. ¹³

A useful working definition, drawn from the selenium-detector literature, keeps the two separate: ⁴

• Lag — carry-over of signal from a previous image into subsequent images. It decays with time and can be seen as a faint positive (or occasionally negative) replica of the prior scene.
• Ghosting — a change in detector sensitivity or gain produced by the exposure history, so a subsequent flat-field exposure is non-uniform in the shape of the earlier high-contrast object, even after any residual signal has decayed.

Because they have different mechanisms, a detector can show one without the other. Early indirect flat-panel imagers were characterized mainly by lag, while some direct selenium mammography systems were dominated by sensitivity-change ghosting. ³⁴

Where these artifacts come from in the imaging chain

Lag and ghosting are receptor-level effects, distinct from the geometric, grid, and exposure factors that dominate other radiographic quality-control tests. They live in the physics of the detector material itself, which is why they are best evaluated at the receptor with the anti-scatter grid and automatic exposure control removed from the question. For the receptor-level tests that surround lag and ghosting — uniformity, signal-to-noise, and defective-pixel counts — see our companion guide on flat-panel detector uniformity and dead-pixel QC, and for the closely related dose-feedback metrics see digital radiography exposure index. ²

Key Technical Principles

The charge-trapping physics

Both effects trace back to charge that is captured by localized trap states and released slowly. The mechanism differs by detector type. ⁵⁶⁷

In indirect-conversion detectors (a cesium iodide or gadolinium oxysulfide scintillator coupled to a hydrogenated amorphous-silicon, a-Si:H, photodiode/thin-film-transistor array), two processes contribute. First, the amorphous-silicon layer contains defect states — dangling bonds and band-tail states in the mobility gap — that capture photo-generated charge and re-emit it with a distribution of time constants; this is the dominant source of electronic lag. ⁶⁷ Second, the scintillator itself exhibits afterglow (phosphorescence), a delayed light emission from charge trapped in the scintillator crystal that adds a second, optical lag component. ⁵

In direct-conversion detectors (an amorphous-selenium, a-Se, photoconductor that converts X-rays straight to charge), deep bulk traps capture carriers and distort the internal electric field. Trapped charge that persists changes the local sensitivity of the photoconductor, which is why selenium systems are prone to sensitivity-change ghosting; hole trapping in particular produces first-frame lag. ⁴⁸

The practical consequence is that lag is not a single number but a decay curve. The first frame after an exposure carries the most residual signal, and successive frames carry progressively less as the traps empty.

How lag is quantified

The standard acceptance measurement acquires a bright, uniform exposure, then reads out one or more frames with no additional radiation. The first-frame lag is the ratio of the residual signal in the first dark frame to the signal just before readout:

$${\text{Lag}}_1=\frac{S_1-S_{\text{dark}}}{S_0-S_{\text{dark}}}\times 100\%$$

where $S_0$ is the mean signal in the last irradiated frame, $S_1$ is the mean residual signal in the first non-irradiated frame, and $S_{\text{dark}}$ is the detector offset (dark) level. Higher-order lag (${\text{Lag}}_2$, ${\text{Lag}}_3$, …) uses the same expression for later frames and characterizes the relaxation tail. ³

For an indirect a-Si:H flat panel, the classic characterization by Siewerdsen and Jaffray reported first-frame lag on the order of a few percent — roughly 2–10% depending on exposure and timing — falling steeply thereafter, to on the order of 0.7%, 0.4%, and 0.3% in the second, third, and fourth frames at moderate sensor loading. Ghost-image contrast could persist for tens of minutes under demanding conditions. ³

How ghosting is quantified

Ghosting is measured as a sensitivity change, not a leftover signal. A high-contrast object (for example, an edge or bar pattern) is imaged to "burn in" an exposure history, and then a uniform "probe" exposure is acquired. The ghosting factor compares the probe signal in a region that was previously heavily exposed with a region that was not:

$$\text{GF}=\frac{S_{\text{fresh}}-S_{\text{ghost}}}{S_{\text{fresh}}}\times 100\%$$

where $S_{\text{fresh}}$ is the probe signal where the detector had little prior exposure and $S_{\text{ghost}}$ is the probe signal where it was pre-exposed. A positive GF means the pre-exposed region has lost sensitivity. In a clinical selenium digital mammography system, lag was measured to be very small (on the order of 0.15% of the source signal) while the sensitivity-change ghosting reached up to about 15% under aggressive conditions — a clear illustration that a detector can be almost lag-free yet still ghost. ⁴ Newer avalanche-selenium research receptors have pushed ghosting below about 3% with first-frame lag near 2%, showing how detector engineering targets these effects directly. ⁸

Comparison of the key temporal metrics

The values in the table are drawn from the peer-reviewed literature to show realistic magnitudes and mechanisms. They are not universal pass/fail tolerances. AAPM TG-150 treats the flat-panel ghost check as a qualitative or optional test (indirect flat panels are much less susceptible than storage-phosphor systems), and acceptance thresholds for detector metrics are generally referenced to the manufacturer specification and the unit's own commissioning baseline rather than a single fixed number. ¹²

Exposure index and deviation index

Lag and ghosting are temporal artifacts, but the everyday DR quality-control metric that a technologist sees is the exposure index (EI) and its companion the deviation index (DI), standardized in IEC 62494-1 and adopted into US practice through AAPM TG-116. The deviation index expresses how far the detector exposure of an image was from the target:

$$\text{DI}=10\cdot {\log }_{10}\left(\frac{\text{EI}}{{\text{EI}}_T}\right)$$

where EI is the measured exposure index and ${\text{EI}}_T$ is the target exposure index. A DI of 0 is on target; because the scale is logarithmic, a DI of +1 corresponds to roughly 25% more detector exposure than intended, and a DI of −1 to roughly 20% less. ¹⁰ TG-116 originally suggested an acceptable band of −1 to +1, but the follow-up survey in AAPM TG-232 found that real clinical DI distributions are broad — often with a standard deviation well above 1 and a mean far from 0 — and recommended that facilities set action limits from their own DI data (for example, at ±1 and ±2 standard deviations) rather than the fixed ±1 band. ¹¹ Consistent EI/DI tracking also surfaces detector-level problems: differences of tens of percent in detector dose have been documented between units nominally calibrated to the same EI. ¹²¹³

Clinical Impact

A detector with excessive lag or ghosting can create structure that is not there, or hide structure that is. A faint residual replica of a prior high-contrast object — a lead marker, a prosthesis edge, a dense bone cortex — can be misread as a line, a device, or a subtle finding on the following patient's image. Conversely, a ghosted region of reduced sensitivity can suppress low-contrast detail exactly where the prior object was imaged. ³⁴

The risk is highest in high-throughput, high-contrast, and rapid-sequence workflows: a busy trauma or portable chest station cycling exposures every few seconds gives the detector less time to relax between images, so residual signal from the previous frame is larger. Because portable and mobile units are frequently the busiest receptors in a hospital, they deserve particular attention; see our guide to mobile radiography radiation safety for the operational context.

Ghosting also interacts with dose. Because ghosting is driven by exposure history, chronic overexposure — the "dose creep" that EI/DI monitoring is designed to catch — both increases patient dose and loads the detector more heavily, deepening any sensitivity-change artifact. Keeping exposures on target therefore protects both the patient and the image. ¹⁰¹¹

Practical Optimization Tips

A practical program for controlling lag and ghosting combines detector characterization at acceptance with disciplined ongoing quality control.

1. Characterize the detector at acceptance

At acceptance and after any detector replacement or major service, a qualified medical physicist should measure first-frame and higher-order lag, perform a ghost/persistence check, and record the results as the unit's baseline. Following an AAPM TG-150 style protocol keeps the method reproducible and the baseline defensible. ¹²

2. Use the manufacturer's offset and gain calibration schedule

Detector offset (dark) and gain (flat-field) calibrations directly affect how residual charge and sensitivity variation are handled. Follow the manufacturer's recommended calibration frequency, and recalibrate after service, because a stale gain map can turn a small sensitivity change into a visible artifact. ¹²

3. Respect the inter-exposure recovery time

Because lag decays over seconds, allowing the detector its specified recovery interval between exposures — especially after a very high-exposure image — reduces carry-over into the next frame. Rapid re-exposure of the same receptor is the single most common way to make lag clinically visible. ³

4. Track exposure index and deviation index

Trend EI and DI by room, body part, and technologist. A drifting mean DI signals technique or calibration problems; a widening DI spread signals inconsistent technique or an emerging detector issue. Set action limits from your own data per TG-232, and investigate outliers. ¹¹¹²

5. Watch for ghosting in the real workflow

A single test image can look clean while a busy clinical sequence ghosts. Periodically inspect flat-field images acquired after high-contrast exposures — the qualitative "passive ghost" check — and be alert to reports of faint repeated structures from radiologists. ¹³

Common pitfalls to avoid

• Confusing lag with ghosting. They have different mechanisms and fixes; a gain recalibration addresses sensitivity-change ghosting but not electronic lag.
• Testing only with a single, well-spaced exposure. This hides the rapid-sequence behavior that matters clinically.
• Ignoring the scintillator. In indirect detectors, afterglow adds an optical lag component that electronics tuning alone will not remove. ⁵
• Treating EI/DI limits as universal. TG-232 specifically recommends site-specific action limits over the fixed ±1 band. ¹¹
• Skipping re-baselining after service. A detector board swap or recalibration changes the temporal behavior; the old baseline no longer applies.

Regulatory Considerations

Digital radiography detectors are radiation-producing-machine equipment regulated at the state level, with quality expectations set by AAPM guidance and accreditation programs rather than by the NRC. Unlike radioactive material, which falls under NRC or Agreement State byproduct-material rules, X-ray systems are regulated by state radiation-control programs (with device requirements informed by the FDA), and their image-quality quality control is shaped primarily by AAPM task-group reports and accreditation standards. ¹²

Key frameworks to reference:

• AAPM Task Group 150 — acceptance testing and quality control performed by the qualified medical physicist for digital radiographic imaging systems, including detector-level tests. ¹
• AAPM Task Group 151 — the ongoing (routine) quality-control program in digital radiography, the technologist-facing companion to TG-150. ¹
• AAPM Task Group 116 and Task Group 232 — the exposure-indicator standard and the current guidance on exposure-index and deviation-index practice. ¹⁰¹¹
• IEC 62494-1 and IEC 62220-1-1 — the international standards defining the exposure index/deviation index and the detective quantum efficiency (DQE) measurement, respectively.

Facilities pursuing or maintaining ACR Digital Radiography accreditation should confirm that the physicist's annual survey and the acceptance report document detector performance, including temporal artifacts where the protocol calls for them, and that ongoing QC and EI/DI tracking are in place. DRPS aligns these evaluations with the facility's state requirements and accreditation program across our service areas. For the broader QC picture, see our guides on ACR accreditation physics requirements and repeat-reject analysis.

Frequently Asked Questions (FAQs)

What is the difference between lag and ghosting in a DR detector?

Lag is residual signal carried over from a previous exposure into later frames — a real signal that decays over time. Ghosting is a change in the detector's sensitivity (gain) caused by its exposure history, so a later uniform exposure reads unevenly even though no leftover signal is present. Lag adds signal; ghosting changes how the detector responds. They come from different physical mechanisms and are measured differently. ¹⁴

What causes image lag in a flat-panel detector?

Lag comes from charge that is temporarily trapped and then slowly released. In indirect (scintillator plus amorphous-silicon) detectors, trap states in the amorphous silicon and afterglow in the cesium iodide scintillator both contribute. In direct amorphous-selenium detectors, deep charge traps in the photoconductor release charge over time. The trapped charge escapes with a spread of time constants, producing residual signal in subsequent frames. ⁵⁶⁷

How is lag measured during acceptance testing?

A physicist acquires a bright exposure, then reads out one or more dark frames with no new radiation. Lag is the residual signal in the first dark frame divided by the pre-readout signal, expressed as a percent. Higher-order lag is measured in the second, third, and later frames to characterize how quickly the detector relaxes back to baseline. ³

Do modern flat-panel detectors still have a ghosting problem?

Modern indirect flat-panel detectors are far less susceptible to visible ghosting than early storage-phosphor and some selenium systems, which is why AAPM TG-150 treats the flat-panel ghost check as an optional or qualitative test rather than a strict numeric tolerance. Ghosting still matters for high-throughput and high-contrast workflows, so a periodic qualitative check remains good practice. ¹³

What is the deviation index and how does it relate to detector QC?

The deviation index reports how far the detector exposure of an image was from its target, on a logarithmic scale: DI equals ten times the base-ten logarithm of the exposure index divided by the target exposure index. A DI of zero is on target, a DI of plus one is roughly 25 percent overexposed, and a DI of minus one is roughly 20 percent underexposed. Tracking DI is part of ongoing DR quality control under AAPM TG-116 and TG-232. ¹⁰¹¹

How often should lag and ghosting be evaluated?

Lag, ghosting, and detector relaxation are typically characterized at acceptance and after major service or detector replacement by a qualified medical physicist under an AAPM TG-150 style protocol. Ongoing quality control, including exposure-index and deviation-index tracking and periodic detector uniformity and artifact checks, follows AAPM TG-151 and the facility's accreditation program on a routine schedule. ¹²

Key Takeaways

• Lag and ghosting are different. Lag is residual signal from a prior exposure; ghosting is a sensitivity/gain change from exposure history. They have different mechanisms, measurements, and fixes. ¹⁴
• Charge trapping is the root cause. Amorphous-silicon trap states, cesium iodide afterglow, and amorphous-selenium deep traps each release charge slowly, producing residual signal (lag) or altered gain (ghosting). ⁵⁶⁷⁸
• Lag is a decay curve, not one number. First-frame lag of a few percent falls steeply over the next frames; characterize the tail, not just the first frame. ³
• Ghosting can occur without lag. Selenium mammography detectors have been nearly lag-free while ghosting up to about 15%. ⁴
• EI/DI tracking is the everyday guardrail. Trend deviation index and set action limits from your own data per TG-232, not a fixed ±1 band. ¹¹
• Baseline at acceptance, then monitor. Characterize under TG-150 at acceptance and after service, and run ongoing QC under TG-151. ¹²

Conclusion

Lag and ghosting are the memory of a flat-panel detector — the faint imprint of the last thing it imaged. They originate in charge trapping in amorphous silicon and selenium and in scintillator afterglow, and they matter most in the busy, high-contrast, rapid-sequence workflows where DR is most valuable. A defensible program characterizes both effects at acceptance against the unit's own baseline, follows the manufacturer's calibration schedule, respects the detector's recovery time, and tracks exposure and deviation indices during routine use. Handled that way, temporal artifacts stay in the physics test report, not on the diagnostic image. ¹²³

How DRPS Can Help

Diagnostic Radiation Physics Services (DRPS) supports imaging facilities with acceptance testing, annual physics surveys, and ongoing quality-control program design for digital radiography systems. For flat-panel detectors, this includes lag and ghosting characterization, detector uniformity and defective-pixel evaluation, exposure-index and deviation-index program setup, and artifact investigation, performed by board-certified medical physicists and aligned with AAPM TG-150/TG-151 and your accreditation program.

DRPS provides diagnostic radiography physics testing, accreditation support, and medical physics consulting across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware.

A strong detector QC program is not about chasing a single test image. It is about knowing how the receptor behaves over a real clinical day.

Related Resources

• Flat-panel detector uniformity & dead-pixel QC
• Digital radiography exposure index
• Detective quantum efficiency in digital radiography
• Repeat-reject analysis
• ACR accreditation physics requirements
• Diagnostic radiography physics testing
• Accreditation support

References

1. Jones AK, Heintz P, Geiser W, Goldman L, Jerjian K, Martin M, 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. doi.org
2. 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. doi.org
3. Siewerdsen JH, Jaffray DA. A ghost story: spatio-temporal response characteristics of an indirect-detection flat-panel imager. Medical Physics. 1999;26(8):1624-1641. doi:10.1118/1.598657. doi.org
4. Bloomquist AK, Yaffe MJ, Mawdsley GE, Hunter DM, Beideck DJ. Lag and ghosting in a clinical flat-panel selenium digital mammography system. Medical Physics. 2006;33(8):2998-3005. doi:10.1118/1.2218315. doi.org
5. Mooslechner M, Mitterlechner B, Weichenberger H, Huber S, Sedlmayer F, Deutschmann H. Analysis of a free-running synchronization artifact correction for MV-imaging with aSi:H flat panels. Medical Physics. 2013;40(3):031906. doi:10.1118/1.4790564. doi.org
6. Starman J, Tognina C, Partain L, Fahrig R. A forward bias method for lag correction of an a-Si flat panel detector. Medical Physics. 2012;39(1):18-27. doi:10.1118/1.3664004. doi.org
7. Mail N, O'Brien P, Pang G. Lag correction model and ghosting analysis for an indirect-conversion flat-panel imager. Journal of Applied Clinical Medical Physics. 2007;8(3):137-146. doi:10.1120/jacmp.v8i3.2483. doi.org
8. Orlik C, Léveillé S, Arnab SM, Howansky AF, et al. Improved temporal performance and optical quantum efficiency of avalanche amorphous selenium for low dose medical imaging. Journal of Medical Imaging. 2024;11(1):013502. doi:10.1117/1.JMI.11.1.013502. doi.org
9. Ren C, Kan S, Huang W, Xi Y, Ji X, Chen Y. Lag-Net: Lag correction for cone-beam CT via a convolutional neural network. Computer Methods and Programs in Biomedicine. 2025;266:108753. doi:10.1016/j.cmpb.2025.108753. doi.org
10. Shepard SJ, Wang J, Flynn M, Gingold E, Goldman L, Krugh K, et al. An exposure indicator for digital radiography: AAPM Task Group 116 (executive summary). Medical Physics. 2009;36(7):2898-2914. doi:10.1118/1.3121505. doi.org
11. Dave JK, Jones AK, Fisher R, Hulme K, Rill L, Zamora D, et al. Current state of practice regarding digital radiography exposure indicators and deviation indices: Report of AAPM Imaging Physics Committee Task Group 232. Medical Physics. 2018;45(11):e1146-e1160. doi:10.1002/mp.13212. doi.org
12. Scott AW, Zhou Y, Allahverdian J, Nute JL, Lee C. Evaluation of digital radiography practice using exposure index tracking. Journal of Applied Clinical Medical Physics. 2016;17(6):343-355. doi:10.1120/jacmp.v17i6.6082. doi.org
13. Cohen MD, Cooper ML, Piersall K, Apgar BK. Quality assurance: using the exposure index and the deviation index to monitor radiation exposure for portable chest radiographs in neonates. Pediatric Radiology. 2011;41(5):592-601. doi:10.1007/s00247-010-1951-9. doi.org