SPECT Scatter Correction: TEW, DEW, and Beyond

Introduction

SPECT scatter correction estimates and removes the Compton-scattered photons that leak into the photopeak energy window, restoring image contrast and making quantitative activity measurements trustworthy. Without it, a SPECT image carries a diffuse haze of mispositioned counts that lowers lesion contrast and inflates apparent activity — a problem that becomes critical the moment SPECT is used to measure, not just to visualize. ^{4, 6}

For decades, SPECT was read qualitatively: is the perfusion defect there or not? In that world, scatter softened contrast but was often tolerable. Quantitative SPECT/CT changed the stakes. When a Lu-177 SPECT scan is used to calculate the absorbed dose a tumor or kidney receives, a scatter-driven bias of tens of percent in the estimated activity propagates straight into the dosimetry. ^{5, 6} Scatter correction moved from a nice-to-have to a required element of a defensible quantitative reconstruction.

This guide explains why scattered photons contaminate the photopeak, how the workhorse energy-window methods — triple-energy-window (TEW) and dual-energy-window (DEW) — estimate and subtract them, where model-based and Monte Carlo methods take over, and how to set up and validate scatter correction in a clinical SPECT/CT program. DRPS supports this work through its PET/CT and nuclear medicine physics and medical physics consulting services across Florida, Maryland, Virginia, Washington DC, California, and Nevada.

Topic Explanation

Why scattered photons corrupt SPECT

A gamma camera forms an image by accepting photons whose measured energy falls within a window centered on the radionuclide's photopeak. The implicit assumption is that a photon detected at a given position traveled in a straight line from its point of emission through the collimator to the detector. Compton scatter breaks that assumption. ⁶

When a photon Compton-scatters in the patient, it changes direction and loses energy. If it is subsequently detected, the camera records it along a line of response that does not point back to where it was emitted. Ideally, the energy loss would push every scattered photon out of the photopeak window and it would simply be rejected. But a gamma camera's energy resolution is finite — typically around 9–10% at 140 keV for a modern NaI(Tl) system — so a scattered photon that lost only a little energy still registers inside the photopeak window. ^{4, 6}

The result is a diffuse, low-frequency background of mispositioned counts superimposed on the true activity distribution. Its practical effects are:

Reduced contrast, because cold regions fill in with scattered counts.
Overestimated activity, because the photopeak counts include scatter that does not belong to the source in that voxel. ⁸
Degraded quantitation, which matters most for SUV-like uptake metrics and for dosimetry. ^{4, 5}

For how scatter correction fits alongside the other two big corrections — attenuation and resolution recovery — see our guides to SPECT reconstruction with FBP and OSEM and quantitative SPECT/CT calibration.

The size of the problem

Scatter is not a small correction. The fraction of photopeak counts that are actually scatter — the scatter fraction — depends on the radionuclide, energy window, collimator, and the size and shape of the patient, but it is routinely a substantial share of the recorded signal. For Tc-99m imaging, the scatter-to-primary count ratio is often on the order of 0.3, corresponding to a scatter fraction of roughly a quarter to a third of the counts in the photopeak; for lower-energy or more complex emitters it can be much larger. ¹¹ For I-123, Monte Carlo work has shown that downscatter — including high-energy photons that scatter down into the photopeak — can contribute on the order of 39% of the counts in the window, producing activity overestimation if left uncorrected. ⁸

Key Technical Principles

The triple-energy-window (TEW) method

The TEW method, introduced by Ogawa and colleagues in 1991, is the most widely implemented scatter-correction technique because it is simple, fast, and requires no model of the patient. ¹ It places two narrow scatter sub-windows immediately adjacent to the main photopeak window — one just below it and one just above it. The count density (counts per keV) in each sub-window estimates the scatter contribution at the two edges of the photopeak, and the scatter inside the photopeak is estimated as the area of the trapezoid joining those two edges.

For each projection pixel, the estimated scatter counts in the main window are:

C_{scat} = (\frac{C _{lower}}{W _{lower}} + \frac{C _{upper}}{W _{upper}}) \cdot \frac{W _{main}}{2}

where $C_{lower}$ and $C_{upper}$ are the counts in the lower and upper scatter sub-windows, and $W_{lower}$ , $W_{upper}$ , and $W_{main}$ are the widths in keV of the lower, upper, and main windows. ^{1, 4} The scatter-corrected (primary) counts are then simply:

C_{prim} = C_{main} - C_{scat}

For a single-gamma radionuclide, where there is no higher-energy emission to scatter down from above, the upper sub-window can be omitted and the trapezoid degenerates to a rectangle: ⁴

C_{scat} = \frac{C _{lower}}{W _{lower}} \cdot W_{main}

Worked TEW example

Consider a Tc-99m SPECT projection acquired with a main photopeak window of 140 keV ± 10% (126–154 keV, so $W_{main} = 28$ keV) and two narrow scatter sub-windows each $W_{lower} = W_{upper} = 7$ keV wide. Suppose a given pixel records $C_{main} = 1000$ counts, with $C_{lower} = 120$ counts in the lower sub-window and $C_{upper} = 40$ counts in the upper sub-window. The estimated scatter is:

C_{scat} = (\frac{120}{7} + \frac{40}{7}) \cdot \frac{28}{2} = (17.1 + 5.7) \cdot 14 \approx 320 counts

so the scatter-corrected primary counts are:

C_{prim} = 1000 - 320 = 680 counts

Here scatter made up about 32% of the recorded photopeak counts in that pixel. Left uncorrected, that pixel would overstate the activity by nearly half ( $1000/680 \approx 1.47$ ). This is the mechanism by which uncorrected scatter biases quantitation and, downstream, dosimetry. Because TEW estimates come from narrow, low-count windows, the estimate itself is noisy, so implementations typically smooth the scatter estimate before subtraction to avoid amplifying noise. ⁶

DEW, model-based, and Monte Carlo methods

TEW is not the only approach. The table below summarizes the main families of SPECT scatter correction.

Method	How scatter is estimated	Strengths	Limitations
Dual-energy window (DEW)	One extra window (e.g., below the photopeak); scatter estimated by a scaling or regression, subtracted from the photopeak ^{2, 3}	Simple; only one added window; long clinical history	Assumes a fixed relationship; can under- or over-correct as scatter shape changes
Dual-photopeak window	Splits the photopeak into two halves and uses their ratio to model scatter within the peak ³	Uses only photopeak counts; no separate scatter window	Radionuclide- and camera-specific calibration required
Triple-energy window (TEW)	Trapezoidal interpolation from two narrow windows either side of the photopeak ¹	Fast, patient-independent, handles up- and downscatter	Noisy estimate from low-count windows; needs smoothing
Effective source scatter estimation (ESSE) / model-based	Scatter modeled from the reconstructed distribution and attenuation map, iterated in reconstruction ^{7, 8}	More accurate; models patient-specific scatter	Computationally heavier; needs an attenuation map
Monte Carlo in reconstruction	Full physics simulation of scatter estimated each iteration ^{9, 10}	Most accurate; handles complex spectra (e.g., Y-90 bremsstrahlung)	Most computationally intensive; needs careful modeling

Model-based and Monte Carlo methods are increasingly embedded directly in iterative reconstruction (OSEM), estimating the scatter contribution at each iteration from the current activity estimate and the CT-derived attenuation map. ^{7, 9} They become essentially mandatory where window methods fail — for example Y-90 bremsstrahlung SPECT, whose broad, continuous energy spectrum has no clean photopeak, so quantitative work uses a single wide acquisition window (on the order of 105–195 keV) with Monte Carlo scatter modeling rather than TEW. ⁹

Clinical Impact

Perfusion and qualitative imaging

Even in qualitative work, scatter correction improves lesion contrast and cold-defect detectability. In cardiac SPECT myocardial perfusion imaging, scatter and attenuation together are classic sources of artifact, and correcting them can change the apparent extent of a defect. Thallium-201, imaged at low energies, has a particularly high scatter contribution — with scatter-to-primary ratios reported near unity in myocardial imaging contexts — which is one reason careful correction matters there. ¹¹ For the broader QC context, see our guide to SPECT/CT quality control.

Quantitative SPECT and theranostic dosimetry

The strongest driver of modern scatter correction is quantitative SPECT/CT for radiopharmaceutical therapy. Lu-177 therapies such as Lu-177 DOTATATE and Lu-177 PSMA are imaged at the 208 keV photopeak (about 10% emission yield), and the joint EANM/MIRD guidance for quantitative Lu-177 SPECT treats scatter correction as a required element of the reconstruction chain, alongside CT-based attenuation correction and resolution recovery. ⁵ The reason is direct: dosimetry converts imaged activity into absorbed dose, so a systematic activity bias from uncorrected scatter becomes a systematic dose error in the treatment record. Phantom studies continue to compare TEW and DEW for exactly these applications, with TEW generally preferred as background scatter increases. ^{10, 12} For how the activity map turns into dose, see our guides to Lu-177 theranostics dosimetry and the MIRD schema for internal dosimetry.

Practical Optimization Tips

A scatter-correction setup and validation generally follows this workflow.

1. Set the energy windows deliberately

Center the photopeak window on the radionuclide's photopeak (140 keV for Tc-99m, 208 keV for Lu-177). ^{5, 14}
Use a photopeak window at least about twice the detector energy resolution so primary counts are not discarded. ⁴
Place TEW scatter sub-windows immediately adjacent to the photopeak window, narrow enough to sample the scatter edges but wide enough to collect usable counts.

2. Match the method to the radionuclide

Single-gamma emitters (e.g., Tc-99m) can use single-sided TEW or DEW. ^{1, 4}
Radionuclides with high-energy emissions that downscatter (e.g., I-123, Lu-177) benefit from full TEW or model-based correction. ^{5, 8}
Broad-spectrum emitters (e.g., Y-90 bremsstrahlung) require model-based or Monte Carlo scatter estimation, not window methods. ⁹

3. Manage the noise of the scatter estimate

Smooth the TEW scatter estimate before subtraction to avoid amplifying noise from the low-count sub-windows. ⁶
Confirm the scatter estimate is subtracted consistently within the iterative reconstruction rather than only post hoc where possible.

4. Validate quantitatively

Verify recovery in a phantom with known activity (e.g., a uniform cylinder and hot/cold inserts) with and without scatter correction.
Confirm the SPECT/CT calibration factor and quantitative accuracy meet program tolerances before clinical dosimetry use. ⁴

Common pitfalls to avoid

Skipping scatter correction for quantitative studies. Attenuation correction alone does not remove scatter and can even worsen quantitation if scatter is ignored. ⁶
Using window methods where they do not apply. Y-90 bremsstrahlung and other broad spectra defeat TEW/DEW. ⁹
Ignoring the noise penalty. Subtracting an unsmoothed, low-count scatter estimate injects noise. ⁶
Copying one radionuclide's windows onto another. Windows and methods are energy- and emission-specific. ⁵
Validating only visually. Quantitative and dosimetry applications require phantom-based quantitative validation, not just a nicer-looking image.

Regulatory Considerations

Scatter correction is a technical component of image quality and quantitation, so it sits inside the SPECT/CT quality-control and quantitative-imaging framework rather than under a single regulation. The governing documents are performance standards and professional guidance rather than a specific CFR clause, but the medical use of the radiopharmaceuticals themselves is federally regulated.

NEMA NU-1-2023 — the current standard for performance measurements of gamma cameras, which defines system spatial resolution with and without scatter and collimator scatter/penetration measurements, providing the vendor-comparable performance basis for a SPECT system. ¹²
MIRD Pamphlet No. 23 — foundational guidance on quantitative SPECT for patient-specific dosimetry, including energy-window design and the role of scatter correction. ⁴
MIRD Pamphlet No. 26 (joint EANM/MIRD) — Lu-177-specific quantitative SPECT guidance that treats scatter correction as required. ⁵
IAEA Human Health Reports No. 9 — a reference for the concepts, requirements, and methods of quantitative nuclear medicine imaging. ¹³
SNMMI procedure standards — provide the clinical energy-window and acquisition guidance for specific studies. ¹⁴

Because SPECT uses byproduct material, the underlying medical use is regulated under 10 CFR Part 35 (or the equivalent Agreement State program), with dose limits under 10 CFR Part 20. Among the states DRPS serves, Florida, Maryland, Virginia, California, and Nevada are NRC Agreement States that license medical use under their own radiation-control rules, while Washington, DC is regulated directly by the NRC. A facility performing quantitative SPECT for dosimetry should document its scatter-correction setup, calibration, and validation as part of its quantitative-imaging quality program. For accreditation support around these programs, see accreditation support.

Frequently Asked Questions (FAQs)

What is scatter correction in SPECT?

Scatter correction is the process of estimating and removing the contribution of Compton-scattered photons that are recorded within the photopeak energy window of a SPECT acquisition. Scattered photons carry mispositioned information, so removing them restores image contrast and improves the accuracy of quantitative activity measurements.

Why do scattered photons end up in the photopeak window?

A gamma camera accepts photons within an energy window around the photopeak, but the detector's finite energy resolution means some Compton-scattered photons — which have lost energy — still register inside that window. Because a scattered photon is detected along a line that does not point back to its origin, it adds a diffuse, mispositioned background that lowers contrast and biases quantitation.

How does the triple-energy-window (TEW) method work?

The TEW method places two narrow scatter sub-windows immediately below and above the main photopeak window. The scatter counts in the photopeak are estimated as the area of a trapezoid formed by the count densities in those two sub-windows, scaled by the photopeak window width, then subtracted pixel by pixel. For a single-gamma radionuclide the upper sub-window can be omitted.

What is the difference between dual-energy-window and triple-energy-window methods?

The dual-energy-window (DEW) method uses one additional window (often below the photopeak) to estimate and subtract scatter, whereas the triple-energy-window (TEW) method uses two narrow windows, one on each side of the photopeak. TEW generally handles both self-scatter and downscatter better, while DEW is simpler; the best choice depends on the radionuclide and the imaging task.

Does scatter correction matter for Lu-177 and theranostic dosimetry?

Yes. Quantitative SPECT/CT for Lu-177, Y-90, and other therapy radionuclides underpins patient-specific dosimetry, and uncorrected scatter can bias the estimated activity by tens of percent. Joint EANM/MIRD guidance treats scatter correction as a required component of a quantitative Lu-177 SPECT reconstruction, alongside attenuation and resolution recovery.

How do I set the energy windows for Tc-99m SPECT?

Tc-99m is imaged at its 140 keV photopeak, commonly with a symmetric 15% or 20% window. Narrow scatter sub-windows for TEW are placed just below and above that photopeak window. The photopeak window should be at least about twice the detector's energy resolution to avoid discarding primary counts, and the exact windows should follow the camera manufacturer and procedure-standard guidance.

Who should set up and validate SPECT scatter correction?

A qualified or board-certified medical physicist should configure the energy windows, validate the scatter-correction method against phantom measurements, and confirm quantitative accuracy as part of the SPECT/CT quality-control program, particularly before the system is used for quantitative or dosimetry applications.

Key Takeaways

Scatter contaminates the photopeak. Finite energy resolution lets Compton-scattered photons into the photopeak window, adding a diffuse, mispositioned background. ⁶
It is not a small effect. Scatter fractions on the order of a quarter to a third are typical for Tc-99m, and downscatter can be much larger for I-123 and other emitters. ^{8, 11}
TEW is the workhorse. Two narrow windows either side of the photopeak give a fast, patient-independent scatter estimate via a trapezoidal interpolation. ¹
Match the method to the radionuclide. DEW and TEW suit clean photopeaks; broad spectra such as Y-90 bremsstrahlung need model-based or Monte Carlo correction. ⁹
Quantitation and dosimetry depend on it. For Lu-177 and other therapies, scatter correction is a required part of a quantitative reconstruction, not an optional filter. ⁵
Validate quantitatively. Confirm activity recovery in a phantom before trusting scatter-corrected numbers for dosimetry. ⁴

Conclusion

Scatter correction is where SPECT stops being a picture and becomes a measurement. The scattered photons that survive the energy window are not random noise; they are a structured, quantifiable bias that lowers contrast and inflates activity in a predictable way. The energy-window methods — TEW and DEW — exploit the shape of the energy spectrum to estimate that bias directly from the data, while model-based and Monte Carlo methods reconstruct it from the physics.

For a facility moving into quantitative SPECT/CT and theranostic dosimetry, scatter correction is not a cosmetic step. It is one of the three corrections — with attenuation and resolution recovery — that separate a number you can defend from a number you cannot. A medical physicist should set the windows deliberately, match the method to the radionuclide, manage the noise of the estimate, and validate the result against known activity before it is used to calculate a patient's dose.

How DRPS Can Help

Diagnostic Radiation Physics Services helps nuclear medicine and theranostic programs build SPECT/CT quantitative-imaging workflows that hold up under scrutiny. This includes energy-window and scatter-correction setup, quantitative calibration and validation, SPECT/CT acceptance and performance testing, and dosimetry-program support through PET/CT and nuclear medicine physics, accreditation support, and medical physics consulting.

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

A quantitative SPECT program is only as trustworthy as its corrections. Getting scatter right is what lets a facility turn images into dose with confidence.

Related Resources

References

Ogawa K, Harata Y, Ichihara T, Kubo A, Hashimoto S. A practical method for position-dependent Compton-scatter correction in single photon emission CT. IEEE Transactions on Medical Imaging. 1991;10(3):408-412. doi:10.1109/42.97591. PubMed
Jaszczak RJ, Greer KL, Floyd CE, Harris CC, Coleman RE. Improved SPECT quantification using compensation for scattered photons. Journal of Nuclear Medicine. 1984;25(8):893-900. PubMed
King MA, Hademenos GJ, Glick SJ. A dual-photopeak window method for scatter correction. Journal of Nuclear Medicine. 1992;33(4):605-612. PubMed
Dewaraja YK, Frey EC, Sgouros G, et al. MIRD Pamphlet No. 23: Quantitative SPECT for patient-specific 3-dimensional dosimetry in internal radionuclide therapy. Journal of Nuclear Medicine. 2012;53(8):1310-1325. doi:10.2967/jnumed.111.100123. PubMed
Ljungberg M, Celler A, Konijnenberg MW, et al. MIRD Pamphlet No. 26: Joint EANM/MIRD guidelines for quantitative 177Lu SPECT applied for dosimetry of radiopharmaceutical therapy. Journal of Nuclear Medicine. 2016;57(1):151-162. doi:10.2967/jnumed.115.159012. PubMed
Hutton BF, Buvat I, Beekman FJ. Review and current status of SPECT scatter correction. Physics in Medicine and Biology. 2011;56(14):R85-R112. doi:10.1088/0031-9155/56/14/R01. PubMed
Welch A, Gullberg GT. Implementation of a model-based nonuniform scatter correction scheme for SPECT. IEEE Transactions on Medical Imaging. 1997;16(6):717-726. doi:10.1109/42.650869. PubMed
Du Y, Tsui BMW, Frey EC. Model-based compensation for quantitative 123I brain SPECT imaging. Physics in Medicine and Biology. 2006;51(5):1269-1282. doi:10.1088/0031-9155/51/5/016. PubMed
Dewaraja YK, Chun SY, Srinivasa RN, et al. Improved quantitative 90Y bremsstrahlung SPECT/CT reconstruction with Monte Carlo scatter modeling. Medical Physics. 2017;44(12):6364-6376. doi:10.1002/mp.12597. PubMed
Michael K, Frangos S, Iakovou I, et al. The impact of dual and triple energy window scatter correction on I-123 postsurgical thyroid SPECT/CT imaging. Life. 2024;14(1):113. doi:10.3390/life14010113. PubMed
Changizi V, Takavar A, Babakhani A, Sohrabi M. Scatter correction for heart SPECT images using TEW method. Journal of Applied Clinical Medical Physics. 2008;9(3):136-140. doi:10.1120/jacmp.v9i3.2767. PubMed
National Electrical Manufacturers Association. NEMA NU 1-2023: Performance Measurements of Gamma Cameras. Rosslyn, VA: NEMA; 2023. nema.org
International Atomic Energy Agency. Quantitative Nuclear Medicine Imaging: Concepts, Requirements and Methods. IAEA Human Health Reports No. 9. Vienna: IAEA; 2014. iaea.org
Society of Nuclear Medicine and Molecular Imaging. SNMMI Procedure Standards. Reston, VA: SNMMI. snmmi.org