Photon-Counting CT: Physics, Image Quality & Dose

Photon-counting detector CT (PCD-CT) replaces the scintillator-and-photodiode chain of conventional CT detectors with direct-conversion semiconductors that count individual X-ray photons and sort them by energy. That single change removes most electronic noise, unlocks ultra-high spatial resolution, delivers spectral data on every scan, and improves dose efficiency — while also giving the medical physicist new things to measure at acceptance and annual testing.¹²⁴

Photon-counting CT is the first fundamental redesign of the CT detector in a generation, and it changes several assumptions that diagnostic physicists have relied on for decades: that electronic noise sets a floor at low dose, that spatial resolution is limited by detector element size and the optical spread in a scintillator, and that spectral imaging requires a dedicated dual-energy mode. This guide explains the underlying physics, the image-quality and dose implications, and — most importantly for a QC program — what changes for acceptance testing, CT-number accuracy, and accreditation.

Introduction

The dominant difference between photon-counting and conventional CT is how the detector turns X-rays into a signal, and that difference propagates into image quality, dose, and quality control. Conventional multidetector CT uses energy-integrating detectors (EIDs): a scintillator converts each X-ray to visible light, a photodiode converts light to charge, and the system integrates the charge over the exposure. A photon-counting detector (PCD) skips the light-conversion step entirely, converting each X-ray directly to a charge pulse in a semiconductor such as cadmium telluride (CdTe) and counting pulses that exceed one or more energy thresholds.¹²⁴

The first clinical photon-counting CT system, the Siemens NAEOTOM Alpha, received FDA clearance in September 2021 (510(k) K211591) — described by the agency as the first major new CT technology in nearly a decade.¹⁵ As of this writing, additional major vendors are actively developing photon-counting platforms, and the technology is moving from research to routine clinical service in cardiac, neuro, musculoskeletal, chest, and abdominal imaging.¹¹¹

For the medical physicist, the takeaway is not just "sharper images." It is that a PCD-CT scanner behaves differently on the QC bench: spatial resolution can exceed what standard phantoms were designed to score, virtual monoenergetic images (VMI) change how HU values are interpreted, and spectral accuracy becomes a testable quantity. DRPS supports facilities adopting these systems through CT physics testing, accreditation support, and medical physicist consulting across Florida, Maryland, Virginia, Washington DC, California, Nevada, Pennsylvania, New York, New Jersey, and Delaware.

Topic Explanation

What is a photon-counting detector?

A photon-counting detector is a direct-conversion semiconductor that produces a measurable charge pulse for each interacting X-ray and counts the pulses whose energy exceeds programmable thresholds. The semiconductor — commonly cadmium telluride (CdTe) or cadmium zinc telluride (CZT) — is held under a strong bias voltage. When an X-ray deposits energy, it liberates electron–hole pairs proportional to the photon energy; the charge cloud drifts to pixel anodes and generates a fast pulse whose height encodes the photon's energy.¹²⁴

Because the system compares each pulse to energy thresholds and increments a counter, two things follow directly:

Electronic (readout) noise is rejected. Baseline electronic noise sits below the lowest energy threshold, so it does not register as counts. In an EID, that same noise is integrated into the signal and matters most at low dose.¹²
Energy information is preserved. Sorting counts into two or more energy bins provides spectral data on every acquisition, without a separate dual-energy scan.¹²⁴

How does this differ from a conventional energy-integrating detector?

In an EID, each detected X-ray is weighted by its energy, and the reflector/septa structure plus optical spreading in the scintillator limits how small the detector elements can be made. A PCD gives every counted photon equal weight (each is "one count"), which improves the contrast of low-energy photons that carry most of the iodine and soft-tissue contrast, and its small sub-pixel structure enables much finer intrinsic spatial sampling.¹³

For background on the conventional metrics these systems still have to satisfy, see our guides to CT image quality: MTF and low-contrast detectability and CT number (HU) calibration QC.

Key Technical Principles

PCD-CT versus EID-CT at a glance

Characteristic	Energy-integrating detector (EID)	Photon-counting detector (PCD)
Conversion pathway	X-ray → light (scintillator) → charge (photodiode)	X-ray → charge directly (CdTe/CZT semiconductor)
Signal formation	Integrates total energy over exposure	Counts individual photons above energy thresholds
Electronic noise	Integrated into signal; limits low-dose performance	Rejected below the lowest threshold
Energy weighting	Higher-energy photons weighted more	Each counted photon weighted equally
Spectral data	Requires a dedicated dual-energy mode/hardware	Inherent on every scan via energy bins
Intrinsic spatial resolution	Limited by element size and optical spread	Sub-pixel elements enable ultra-high resolution
Reported in-plane resolution	Task- and system-dependent	Down to ~0.125–0.2 mm on first clinical system ⁴¹³

The values in the table are representative of published performance on the first clinical PCD-CT system and research prototypes; actual performance is system-, protocol-, and reconstruction-dependent and should be measured locally.³⁴¹³

Why counting photons improves the signal-to-noise trade-off

Image noise in CT is quantum-limited when electronic noise is negligible, so the noise standard deviation scales inversely with the square root of the number of detected photons — and therefore with the square root of dose:

σ \propto \frac{1}{N} \propto \frac{1}{D}

Contrast-to-noise ratio (CNR) for two regions with mean CT numbers $μ_{A}$ and $μ_{B}$ against background noise $σ$ is:

CNR = \frac{∣ μ _{A} - μ _{B} ∣}{σ}

To compare systems fairly at different doses, physicists use a dose-normalized CNR:

CNRD = \frac{CNR}{D}

A PCD improves CNRD in two ways: by removing the electronic-noise contribution to $σ$ (especially important for large patients and low-dose exams) and by equally weighting the low-energy photons that carry iodine contrast.²⁶

Worked example: turning a CNR gain into a dose reduction

Because noise scales as $1/ D$ , the dose needed to reach a fixed image-quality target scales with the square of the required noise. If a PCD improves iodine CNR by a factor $g$ at equal dose, the same original CNR can be achieved at a reduced dose:

\frac{D _{PCD}}{D _{EID}} = (\frac{1}{g})^{2}

Sawall and colleagues reported iodine contrast-to-noise improvements up to about 37% with optimized energy-bin weighting, i.e., $g \approx 1.37$ .⁶ Holding image quality fixed, the corresponding dose ratio is:

\frac{D _{PCD}}{D _{EID}} = (\frac{1}{1.37})^{2} \approx 0.53

That is roughly a 47% dose reduction for the iodine-detection task — consistent with the up-to-46% potential dose reduction that same study reported.⁶ This is a task-specific illustration, not a global guarantee: the achievable reduction depends on the clinical task, patient size, protocol, and reconstruction, and must be validated per protocol before it is claimed clinically.

Ultra-high spatial resolution and the sampling limit

The Nyquist-limited spatial frequency for a detector with effective (isocenter-projected) element pitch $Δ$ is:

f_{Nyquist} = \frac{1}{2 Δ}

PCD sub-pixels reduce $Δ$ well below conventional element sizes, enabling ultra-high-resolution modes. Research PCD-CT has demonstrated up to about 33 line pairs per centimeter, and the first clinical system provides maximum in-plane resolution on the order of 0.125–0.2 mm.³⁴¹³ Focal-spot size, geometry, and reconstruction still constrain the delivered resolution, so measured task transfer functions — not detector geometry alone — should define the achievable resolution in QC.¹⁴

Inherent spectral imaging

Because energy bins are available on every scan, PCD-CT reconstructs virtual monoenergetic images (VMI), iodine maps, virtual non-contrast images, and other material-specific images from a single acquisition. Attenuation is modeled as a weighted sum of basis functions (for example, photoelectric and Compton, or two basis materials):

μ (E) = a_{1} f_{1} (E) + a_{2} f_{2} (E)

Solving for the coefficients $a_{1}$ and $a_{2}$ from the measured energy bins yields the spectral image series.¹² A practical QC implication follows from VMI: HU values depend on the selected keV. Published work indicates a virtual monoenergetic level near 70 keV approximates the attenuation of a conventional 120 kV polychromatic image, which matters when comparing CT numbers against historical baselines.⁹

Clinical Impact

Photon-counting CT changes clinical practice most where conventional CT is limited by resolution, low-dose noise, or the need for spectral characterization. Reported and emerging benefits include:

Cardiac and coronary imaging: higher spatial resolution reduces calcium and stent blooming, improving lumen assessment; dual-source PCD geometry provides temporal resolution on the order of 66 ms.⁴⁷¹⁰
Musculoskeletal and temporal-bone imaging: ultra-high-resolution modes improve visualization of trabecular bone and fine temporal-bone structures.³¹³
Chest and lung imaging: high resolution improves small-airway and interstitial assessment at competitive dose.
Iodine and contrast optimization: improved iodine CNR supports reduced contrast-media volume or reduced dose for a given task.⁶
Quantitative imaging: inherent spectral data support quantitative measures such as iodine concentration and extracellular volume, advancing quantitative CT.⁸¹²

These benefits are realized only when protocols, reconstructions, and reader workflows are built around them — which is why protocol development on PCD-CT is best treated as a collaborative effort among the radiologist, technologist, and medical physicist.¹¹ The same protocol-optimization discipline that governs conventional CT applies here; see CT protocol optimization and metal artifact reduction in CT.

Practical Optimization Tips

Treat acceptance and commissioning as task-based, not just pass/fail

Standard CT QC — CT number accuracy, uniformity, noise, and slice thickness — still applies. But ultra-high-resolution and spectral capabilities are poorly captured by legacy pass/fail phantom criteria alone. Task-based metrics from AAPM Report No. 233 characterize the system more completely:¹⁴

Task transfer function (TTF) for resolution as a function of contrast and dose, rather than a single limiting-resolution value.
Noise power spectrum (NPS) for noise texture, which changes with iterative and deep-learning reconstruction.
Detectability index (d′) for a defined imaging task.

Verify CT-number and spectral accuracy across keV

On PCD-CT, CT numbers depend on the reconstructed image type and VMI energy. During commissioning:

Confirm HU accuracy on the conventional (polychromatic-equivalent) series against your historical baseline, noting that ~70 keV VMI approximates 120 kV HU.⁹
Verify iodine quantification accuracy using a known-concentration phantom; the first clinical system reported iodine CT-number errors of a few percent in phantom testing.⁴
Document which series (energy, reconstruction kernel, denoising) is used for each clinical and QC task.

Manage the resolution–noise–dose triangle deliberately

Ultra-high-resolution modes increase noise for a given dose because smaller voxels contain fewer photons. Decide, per protocol, whether the clinical task needs maximum resolution or whether a standard-resolution reconstruction at lower noise is preferable. Reconstruct at the voxel size the task requires — not the maximum the scanner allows.³⁷

Coordinate with reconstruction and denoising

Deep-learning and iterative reconstruction interact strongly with PCD data and change noise texture and low-contrast detectability. Evaluate the specific reconstruction settings used clinically, and re-evaluate after software upgrades, because a reconstruction change can alter measured performance as much as a hardware change.¹³¹⁴

Common pitfalls to avoid

Assuming dose reduction is automatic. Dose-efficiency gains are task-specific and must be validated per protocol before clinical claims are made.⁶¹¹
Scoring UHR data with legacy assumptions. Limiting-resolution line-pair scoring may not reflect true task performance; use TTF/NPS where possible.¹⁴
Ignoring VMI energy when reading CT numbers. HU depends on keV; comparisons to historical baselines must account for the selected energy.⁹
Skipping post-upgrade re-testing. Reconstruction and firmware updates can shift measured performance.¹⁴

Regulatory Considerations

A photon-counting CT scanner is a radiation-producing machine, so it is regulated as CT — under FDA and state radiation-control authority — not under NRC byproduct-material rules. The clinical device is marketed under FDA 510(k) clearance (for example, K211591 for the first clinical system) and must meet the federal performance standard for CT equipment at 21 CFR 1020.33.¹⁵

Key frameworks:

FDA 510(k) clearance and 21 CFR 1020.33. The scanner must be an FDA-cleared device meeting the CT performance standard, including dose-information reporting (CTDIvol, DLP) at the console.
State radiation-control registration and inspection. X-ray-producing machines are registered and inspected under state radiation-control programs. Every state DRPS serves regulates diagnostic X-ray machines at the state level, in addition to FDA device regulation; there is no NRC medical-use license for a CT scanner because it emits no byproduct material.
ACR CT Accreditation Program. Clinical CT scanners, including PCD-CT, are accredited under the ACR CT program, which requires an annual medical physicist system-performance evaluation, clinical image review, and phantom testing. The program continues to evolve; facilities should confirm current phantom-image submission timelines and any module-specific requirements when they accredit or renew.
AAPM guidance for performance evaluation. AAPM Report No. 233 provides the task-based methodology many physicists use for modern CT commissioning and characterization.¹⁴

Because dose metrics (CTDIvol, DLP, and size-specific dose estimate) still drive protocol review and diagnostic reference level comparisons, the arrival of PCD-CT does not change the physicist's dose-optimization responsibilities; see CTDIvol and DLP dose metrics and size-specific dose estimate (SSDE) in CT. Confirm requirements with the authority having jurisdiction before relying on any single assumption.

Frequently Asked Questions (FAQs)

What is photon-counting CT?

Photon-counting CT (PCD-CT) is a CT technology that uses direct-conversion semiconductor detectors, such as cadmium telluride, to count individual X-ray photons and measure the energy of each one. This differs from conventional energy-integrating detectors, which convert X-rays to light in a scintillator and record only the total energy deposited. Counting photons directly removes most electronic noise and provides spectral information on every scan.

How is a photon-counting detector different from an energy-integrating detector?

An energy-integrating detector (EID) converts each X-ray into visible light in a scintillator, then into an electrical signal, and integrates that signal over the exposure — so a low-energy photon contributes less than a high-energy one and electronic noise is included. A photon-counting detector (PCD) converts the X-ray directly into an electrical pulse in a semiconductor and counts each pulse above an energy threshold, rejecting electronic noise below that threshold and sorting photons into energy bins.

Does photon-counting CT reduce radiation dose?

It can. By removing electronic noise and using detected photons more efficiently, photon-counting CT can produce equivalent image quality at lower dose, or better image quality at the same dose. Peer-reviewed studies have reported iodine contrast-to-noise improvements and potential dose reductions on the order of tens of percent for specific tasks, but the achievable benefit depends on the clinical task, protocol, and reconstruction. Dose reduction is not automatic and must be validated per protocol.

Is photon-counting CT the same as dual-energy CT?

No, although they overlap. Dual-energy CT obtains two energy measurements using two tube potentials, fast kV switching, dual sources, or dual-layer detectors. A photon-counting detector separates energies within a single detector using multiple energy thresholds, so spectral data are available on essentially every acquisition without a separate dual-energy mode, and typically with better spectral separation and spatial resolution.

Does photon-counting CT need special QC or acceptance testing?

Yes. The core CT performance metrics still apply — CT number accuracy, uniformity, noise, and spatial resolution — but the physicist must also evaluate energy-bin behavior, virtual monoenergetic image accuracy, and material-specific measurements. Task-based methods such as those in AAPM Report No. 233 (task transfer function, noise power spectrum, and detectability) are well suited to characterizing ultra-high-resolution and spectral performance during commissioning.

Is photon-counting CT covered by ACR CT accreditation?

Yes. A photon-counting CT scanner used clinically is accredited under the same ACR CT Accreditation Program as any other CT system, including the annual medical physicist system-performance evaluation, clinical image review, and phantom testing. The physicist should confirm that the reconstruction and series submitted for phantom scoring represent the clinical protocols.

Key Takeaways

PCD-CT counts photons directly. Direct-conversion semiconductors (CdTe/CZT) count individual X-rays and sort them by energy, replacing the scintillator–photodiode chain of energy-integrating detectors.¹²
Electronic noise is largely removed. Rejecting sub-threshold noise improves low-dose and large-patient imaging where EIDs are most limited.²
Spectral data come free on every scan. Energy bins enable VMI, iodine maps, and virtual non-contrast images without a dedicated dual-energy mode.¹⁴
Dose efficiency improves, but is task-specific. Reported iodine CNR gains translate to potential dose reductions of tens of percent, which must be validated per protocol.⁶
QC changes. CT-number and spectral accuracy across keV, ultra-high-resolution characterization, and task-based metrics (AAPM 233) belong in acceptance and annual testing.⁹¹⁴
Regulation is CT regulation. FDA 510(k) and 21 CFR 1020.33, state machine registration, and ACR CT accreditation apply; there is no NRC license because CT uses no byproduct material.¹⁵

Conclusion

Photon-counting CT is a genuine step change in CT detector physics, not an incremental upgrade. By counting individual photons and measuring their energy, it removes electronic noise, enables ultra-high spatial resolution, and delivers spectral data on every scan — improvements that translate into better cardiac, neuro, musculoskeletal, and quantitative imaging, and into real, task-specific dose-efficiency gains. For the medical physicist, the technology raises the bar on acceptance testing and QC: CT-number and spectral accuracy across keV, task-based resolution and noise characterization, and careful validation of any dose-reduction claim. Facilities that pair PCD-CT with a disciplined, task-based physics program will capture the technology's benefits while keeping their CT service accreditable and defensible.

How DRPS Can Help

Diagnostic Radiation Physics Services (DRPS) supports facilities evaluating, commissioning, and operating photon-counting and conventional CT systems with CT physics testing, acceptance testing, task-based image-quality characterization, protocol and dose optimization, ACR accreditation support, and medical physicist consulting delivered by board-certified medical physicists.

DRPS serves imaging facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware.

A strong CT physics program is not just about passing accreditation. It is about making sure new detector technology delivers real diagnostic benefit at the lowest reasonable dose — and that the improvement is measured, documented, and defensible.

Related Resources

References

Willemink MJ, Persson M, Pourmorteza A, et al. Photon-counting CT: technical principles and clinical prospects. Radiology. 2018;289(2):293-312. doi:10.1148/radiol.2018172656. PubMed
Flohr T, Petersilka M, Henning A, et al. Photon-counting CT review. Physica Medica. 2020;79:126-136. doi:10.1016/j.ejmp.2020.10.030. PubMed
Leng S, Bruesewitz M, Tao S, et al. Photon-counting detector CT: system design and clinical applications of an emerging technology. RadioGraphics. 2019;39(3):729-743. doi:10.1148/rg.2019180115. PubMed
Rajendran K, Petersilka M, Henning A, et al. First clinical photon-counting detector CT system: technical evaluation. Radiology. 2022;303(1):130-138. doi:10.1148/radiol.212579. PubMed
Zhou W, Bartlett DJ, Diehn FE, et al. Reduction of metal artifacts and improvement in dose efficiency using photon-counting detector CT and tin filtration. Investigative Radiology. 2019;54(4):204-211. doi:10.1097/RLI.0000000000000535. PubMed
Sawall S, Klein L, Amato C, et al. Iodine contrast-to-noise ratio improvement at unit dose and contrast media volume reduction in whole-body photon-counting CT. European Journal of Radiology. 2020;126:108909. doi:10.1016/j.ejrad.2020.108909. PubMed
Symons R, De Bruecker Y, Roosen J, et al. Quarter-millimeter spectral coronary stent imaging with photon-counting CT: initial experience. Journal of Cardiovascular Computed Tomography. 2018;12(6):509-515. doi:10.1016/j.jcct.2018.10.008. PubMed
Mergen V, Sartoretti T, Klotz E, et al. Extracellular volume quantification with cardiac late enhancement scanning using dual-source photon-counting detector CT. Investigative Radiology. 2022;57(6):406-411. doi:10.1097/RLI.0000000000000851. PubMed
Mergen V, Ried E, Allmendinger T, et al. Epicardial adipose tissue attenuation and fat attenuation index: phantom study and in vivo measurements with photon-counting detector CT. American Journal of Roentgenology. 2022;218(5):822-829. doi:10.2214/AJR.21.26930. PubMed
Dane B, Froemming A, Schwartz FR, et al. Photon counting CT clinical adoption, integration, and workflow. Abdominal Radiology. 2024;49(12):4600-4609. doi:10.1007/s00261-024-04503-5. PubMed
Treb KJ, El Sadaney AO, Ferrero A, et al. Quantitative CT imaging: where are we, and what is missing? British Journal of Radiology. 2025;98(1175):1834-1842. doi:10.1093/bjr/tqaf075. PubMed
Chang S, Benson JC, Lane JI, et al. Ultra-high-resolution photon-counting-detector CT with a dedicated denoising CNN for enhanced temporal bone imaging. American Journal of Neuroradiology. 2025;46(6):1188-1195. doi:10.3174/ajnr.A8572. PubMed
Samei E, Bakalyar D, Boedeker KL, et al. Performance evaluation of computed tomography systems: summary of AAPM Task Group 233. Medical Physics. 2019;46(11):e735-e756. doi:10.1002/mp.13763. PubMed
American Association of Physicists in Medicine. Performance Evaluation of Computed Tomography Systems. AAPM Report No. 233. College Park, MD: AAPM; 2019. aapm.org
U.S. Food and Drug Administration. September 2021 510(k) Clearances (NAEOTOM Alpha, K211591). fda.gov
American College of Radiology. CT Accreditation Program. acr.org