Partial Volume Effect in Quantitative SPECT
The partial volume effect is the loss of measured signal that occurs when a structure is small relative to the SPECT system's spatial resolution, and it is the single largest source of bias in quantifying small lesions and organs. Limited resolution blurs counts outward across an object's edge (spill-out) and blurs surrounding activity inward (spill-in), so a small lesion's measured activity concentration can read at a fraction of its true value. Recovery coefficients—ratios of measured to true concentration, measured from spheres of known activity—quantify that loss and correct it so activity, SUV, and therapy dosimetry stay accurate.12
Quantitative SPECT/CT has moved from a research curiosity to a clinical tool for theranostics, bone imaging, and cardiac work, and its accuracy now feeds patient-specific dosimetry. A system can correct attenuation, scatter, and detector response beautifully and still report the wrong number for a 1 cm tumor if the partial volume effect is ignored.23 This article explains what the effect is, how recovery coefficients are measured and applied, where it bites hardest clinically, and how to build a defensible partial volume workflow.
Introduction
Absolute quantification in SPECT converts reconstructed counts into an activity concentration in becquerels per milliliter, and every step in that chain has to be right for the final number to mean anything. The chain includes a traceable system calibration factor, attenuation correction from the CT, scatter correction, collimator-detector response modeling, and—critically for small objects—partial volume correction.23 Of these, the partial volume effect is unique because it is a direct, physical consequence of finite spatial resolution that no amount of calibration can remove on its own.
The reason the effect has become a front-line concern is the rise of radiopharmaceutical therapy. When the goal was a qualitative image, a slightly blurred small lesion was a cosmetic issue. Now that the same images are used to estimate absorbed dose for Lu-177 DOTATATE, Lu-177 PSMA, and similar therapies, a factor-of-two bias in a small-lesion measurement becomes a factor-of-two bias in the dose that is supposed to guide treatment.24 DRPS supports facilities building or validating this capability through its PET/CT and nuclear medicine physics services across Florida, Maryland, Virginia, Washington DC, California, Nevada, and the rest of our service area.
This is a companion to our broader guide on quantitative SPECT/CT calibration: calibration gets the global scale right, while partial volume correction gets small objects right.
Topic Explanation
What the partial volume effect actually is
The partial volume effect is the discrepancy between the true activity distribution and what the imaging system reports, caused by the system's finite spatial resolution and by finite voxel sampling.1 It has two physical components that are easy to confuse and important to separate:
- Spill-out: counts that physically originate inside the object are blurred outward beyond its edges because the system's point spread function has a finite width. The object therefore appears to contain fewer counts than it really does.
- Spill-in: counts that originate in surrounding tissue blur inward and are mistakenly attributed to the object. In a hot lesion on a cold background, spill-out dominates and the lesion reads low; in a cold lesion on a hot background, spill-in dominates and the lesion reads falsely high.1
A second, smaller contributor is tissue-fraction effect, or voxel sampling: a voxel straddling the boundary of the object contains a mixture of object and background, and the reconstruction assigns it a single blended value. In practice, for nuclear medicine the resolution-driven blurring dominates.1
The effect is not a reconstruction bug and not a calibration error. It is physics: a gamma camera with a parallel-hole collimator has a spatial resolution—expressed as the full width at half maximum (FWHM) of its point spread function—that is typically on the order of 7 to 15 mm at clinically relevant distances, depending on collimator, radionuclide, and radius of rotation. Any object comparable to or smaller than that resolution cannot be faithfully recovered without correction.13
Why quantitative SPECT makes it unavoidable
In qualitative imaging, the reader compensates mentally for blurring. In quantitative imaging, a number is extracted from a volume of interest and used downstream—as an SUV, as an input to a dosimetry calculation, or as a longitudinal biomarker of response. The partial volume effect systematically biases that number, and because the bias depends on object size, it distorts comparisons between large and small structures and between the same lesion imaged before and after it shrinks with therapy.14
The MIRD committee's guidance on quantitative SPECT for dosimetry lists partial volume correction alongside attenuation, scatter, and resolution recovery as a required element of a defensible quantification chain, and the joint EANM/MIRD Lu-177 guidance reiterates the point for therapy dosimetry specifically.25 The EANM practice guideline for quantitative SPECT-CT similarly treats recovery coefficients as part of standard setup and harmonization.3
Key Technical Principles
The recovery coefficient
The recovery coefficient (RC) is the workhorse metric for the partial volume effect. It is the ratio of the measured activity concentration to the true activity concentration for an object of a given size:
An RC of 1.0 means full recovery (no measurable partial volume loss); an RC of 0.5 means the system reports only half of the true concentration. Once the RC for a given object size is known, the partial-volume-corrected concentration is simply:
Recovery coefficients are measured empirically by filling spheres of known volume and known activity concentration—most commonly the six fillable spheres of the NEMA IEC body phantom—imaging them with the clinical protocol, and plotting RC against sphere size. The result is a recovery curve that rises from low values for small spheres toward a plateau near unity for large spheres.368
The underlying model is that the imaging system acts approximately as a convolution of the true activity distribution with the system point spread function. For a spherical object of diameter
Representative recovery coefficients
The table below gives representative, order-of-magnitude recovery coefficients for hot spheres in a cold or low background measured on a modern SPECT/CT system with attenuation, scatter, and collimator-detector response corrections applied. These are illustrative values for teaching, not a substitute for a curve measured on your own system—the exact numbers depend heavily on the radionuclide, reconstruction, and VOI method.68
| Sphere inner diameter | Size vs. system resolution | Representative RC (no partial volume correction) | Practical consequence |
|---|---|---|---|
| ~10 mm | ≤ 1 × FWHM | ~0.2–0.3 | Severe underestimate; activity off by 3–5× |
| ~17 mm | ~1.5 × FWHM | ~0.3–0.5 | Large underestimate; roughly half-truth |
| ~22 mm | ~2 × FWHM | ~0.4–0.6 | Moderate underestimate |
| ~28 mm | ~2.5 × FWHM | ~0.6–0.7 | Appreciable but shrinking bias |
| ~37 mm | ~3.5 × FWHM | ~0.7–0.85 | Minor underestimate |
| ≥ 60 mm | ≫ FWHM | ~0.9–1.0 | Near-complete recovery |
These magnitudes are consistent with published Lu-177 SPECT work. In a Monte Carlo and clinical study of Lu-177 PSMA SPECT, spheres showed recovery coefficients of about 0.26 for a 20 mm sphere and 0.70 for a 40 mm sphere before correction, improving substantially after partial volume correction.8 A phantom study of Lu-177 PRRT reported that applying recovery-coefficient correction improved the average activity-quantification error across inserts from about −26% to roughly +1%.6
A worked example
Consider a SPECT/CT system whose effective reconstructed resolution corresponds to an FWHM of about 10 mm, imaging a 15 mm spherical lesion. From a recovery curve like the one above, the RC for a 15 mm object (about 1.5 × FWHM) is roughly 0.45. Suppose the measured mean activity concentration in the lesion VOI is:
The partial-volume-corrected concentration is:
Uncorrected, this lesion's activity—and therefore its calculated absorbed dose—would be under-reported by more than a factor of two. If this lesion were a critical organ substructure or a small metastasis being tracked for dose-response, that error would dominate every downstream conclusion.48
Resolution recovery helps but does not finish the job
Modeling the collimator-detector response (CDR) inside an iterative reconstruction—often called resolution recovery—partially compensates for blurring and improves recovery coefficients, especially when the model is accurate.912 But it has limits. Residual partial volume effect remains for the smallest objects, and aggressive resolution recovery can introduce edge overshoot and Gibbs-type ringing that distorts quantification near boundaries.8 For this reason, recovery coefficients are still measured after resolution recovery is enabled, and a dedicated partial volume correction (recovery-coefficient scaling, reblurred Van Cittert, or iterative Yang methods) is applied on top when small-object accuracy matters.8
Clinical Impact
Theranostics and dosimetry
The highest-stakes application is radiopharmaceutical therapy dosimetry. Absorbed dose is proportional to the time-integrated activity in the target, so a recovery coefficient of 0.5 on a small lesion propagates almost directly into a 50% underestimate of tumor dose. Studies of Lu-177 therapies have shown that partial volume correction materially changes the activity—and therefore the dose—estimated for small tumors and for organs at risk such as the kidneys, and that correction methods converge the measured values toward the truth.468 For dose-response modeling, where the whole point is to relate delivered dose to outcome, uncorrected partial volume bias can obscure a real relationship entirely.
Bone SPECT/CT SUV
Quantitative bone SPECT/CT now reports SUV to help distinguish benign from malignant vertebral lesions. Because many of these lesions are small, partial volume effect lowers the measured SUV, and correction shifts the diagnostic thresholds. In one bone SPECT/CT series, applying recovery-coefficient correction reduced the SUV cutoff that separated metastasis from degenerative change, underscoring that any quantitative threshold is only valid for the correction state it was derived under.11
Cardiac perfusion and small structures
In cardiac SPECT, the myocardial wall is thin relative to system resolution, so partial volume effect—compounded by cardiac and respiratory motion—limits absolute quantification of regional uptake and of myocardial blood flow. Dedicated correction methods that incorporate higher-resolution anatomical information can reduce the impact and improve the uniformity and count accuracy of the measured wall.10 The lesson generalizes: whenever the structure of interest is thin or small, the measured concentration is suspect until the partial volume question is answered.
Longitudinal and multi-center comparisons
Because the bias is size-dependent, it corrupts comparisons. A lesion that shrinks under therapy will show an apparent drop in concentration partly caused by worsening partial volume effect, not only by real biological change. And two sites with different collimators, protocols, or reconstruction settings will have different recovery curves, so uncorrected SUVs are not comparable between them—one reason harmonization guidance emphasizes measured recovery coefficients.37
Practical Optimization Tips
A defensible partial volume workflow generally follows these steps.
1. Fix and document the acquisition and reconstruction
Lock down the radionuclide, collimator, energy windows, matrix, radius of rotation, number of projections, and the full reconstruction recipe (algorithm, iterations, subsets, post-filter, and which corrections are on). The recovery curve is only valid for this exact configuration.37
2. Measure a recovery curve on your own system
Fill the NEMA IEC body phantom spheres with a known activity concentration, image with the clinical protocol, and compute RC for each sphere using a defined VOI method. Build hot-in-cold and, where relevant, hot-in-warm (realistic contrast) curves, because background changes the result.68
3. Match the VOI method to the calibration
Draw clinical VOIs the same way you drew them to build the recovery curve—same segmentation threshold, same margin. Some workflows deliberately use a VOI larger than the physical object to capture spilled-out counts, pairing it with a recovery coefficient defined for that geometry; what matters is consistency between measurement and calibration.7
4. Report whether correction was applied
Any quantitative value—activity, SUV, absorbed dose—should state whether partial volume correction was used and by what method. A threshold or dose number without that context cannot be reproduced or compared.3
5. Verify with an independent object
Validate the whole chain by measuring an object of known activity that was not used to build the curve, such as an organ-shaped insert or an anthropomorphic phantom, and confirm the corrected value lands within an acceptable tolerance.67
Common pitfalls to avoid
- Borrowing a recovery curve from another scanner, collimator, or protocol.
- Assuming resolution recovery is enough and skipping an explicit partial volume correction for small objects.8
- Changing the VOI method between calibration and clinical measurement.
- Ignoring background contrast, which shifts recovery coefficients through spill-in.1
- Applying correction silently, so downstream users cannot tell corrected from uncorrected numbers.
- Over-correcting noisy small objects, amplifying statistical noise and edge artifacts.8
Regulatory Considerations
Quantitative SPECT used for therapy dosimetry sits at the intersection of medical-physics best practice and the radioactive-material rules that govern medical use. The quantification methods themselves are governed by professional guidance rather than a single regulation, but the clinical context is regulated.
- MIRD Pamphlet No. 23 establishes the general framework for quantitative SPECT for patient-specific 3D dosimetry, including the role of partial volume correction.2
- MIRD Pamphlet No. 26 (joint EANM/MIRD) gives radionuclide-specific guidance for quantitative Lu-177 SPECT applied to dosimetry.5
- The EANM practice guideline for quantitative SPECT-CT provides setup, harmonization, and clinical-use recommendations, including recovery coefficients.3
- 10 CFR Part 35 governs the medical use of byproduct material in the United States, including authorized uses, written directives, and the responsibilities of the authorized user and medical physicist for therapies whose dosimetry these methods support.2
In the United States, radioactive material is regulated by the NRC or by an Agreement State. Of the states DRPS serves, Florida, Maryland, Virginia, California, Nevada, Pennsylvania, New York, and New Jersey are NRC Agreement States that license medical use under their own radiation-control programs, while Washington, DC and Delaware are regulated directly by the NRC. A facility standing up quantitative SPECT for dosimetry should confirm which authority issues its license and coordinate the physics program with its medical physics consulting and radioactive-material license support.
Frequently Asked Questions (FAQs)
What is the partial volume effect in SPECT?
The partial volume effect is the loss of measured signal that happens when an object is small relative to the scanner's spatial resolution. Counts that belong to the object are blurred outward across its edge (spill-out) and counts from nearby activity blur inward (spill-in), so the measured activity concentration in a small lesion is lower than its true value. It is the dominant quantification error for objects smaller than about two to three times the system resolution.
What is a recovery coefficient?
A recovery coefficient is the ratio of the measured activity concentration to the true activity concentration for an object of a given size. It is determined by imaging spheres of known activity in a phantom, such as the NEMA IEC body phantom, and plotting recovery against sphere size. A recovery coefficient of 0.5 means the measured value is only half the truth, so dividing the measured concentration by 0.5 restores the estimate.
Why does the partial volume effect matter for radiopharmaceutical therapy dosimetry?
Absorbed dose scales with the activity in the target, so if the measured activity in a small tumor or at-risk organ is biased low, the calculated absorbed dose is biased low by a similar factor. Without partial volume correction, dosimetry for small lesions can under-report dose by a factor of two or more, which undermines dose-response analysis and personalized treatment planning for therapies such as Lu-177 DOTATATE and Lu-177 PSMA.
How large must a lesion be to avoid significant partial volume effect?
As a rule of thumb, an object needs a diameter of roughly two to three times the reconstructed spatial resolution (full width at half maximum) to recover most of its true activity concentration. For a SPECT system with an effective resolution near 10 to 12 mm, structures smaller than about 20 to 30 mm are meaningfully affected and should be corrected or interpreted with caution.
Does collimator-detector response modeling in reconstruction remove the partial volume effect?
Resolution recovery helps but does not eliminate it. Modeling the collimator-detector response during iterative reconstruction improves recovery and reduces blurring, but residual partial volume effect remains for small objects, and resolution recovery can introduce edge artifacts such as Gibbs ringing. A measured recovery coefficient or a dedicated partial volume correction is still needed for accurate small-object quantification.
What inputs does a recovery coefficient curve depend on?
A recovery coefficient curve is specific to the radionuclide, collimator, acquisition protocol, reconstruction algorithm and settings (iterations, subsets, post-filter), the corrections applied (attenuation, scatter, resolution recovery), the object shape, the background activity, and how the volume of interest is drawn. Because of this dependence, recovery curves should be generated on the same system and protocol used clinically, not borrowed from another site.
Should the volume of interest be drawn tightly around the object?
Not necessarily. For a mean-concentration measurement corrected by a recovery coefficient, the volume of interest should match the segmentation method used to build the recovery curve. Some workflows deliberately use a larger volume of interest that captures the spilled-out counts, with the recovery coefficient defined for that geometry; the key is that measurement and calibration use a consistent definition.
Key Takeaways
- The partial volume effect is physics, not a bug. Finite spatial resolution spills counts out of small objects and in from their surroundings, biasing the measured concentration.1
- Recovery coefficients quantify and correct it. RC is the ratio of measured to true concentration; dividing by RC restores the estimate.36
- Size relative to resolution is what matters. Objects smaller than about two to three times the system FWHM need correction.1
- Dosimetry is the high-stakes case. Uncorrected small-lesion activity can under-report absorbed dose by a factor of two or more in Lu-177 and similar therapies.48
- Resolution recovery is not a complete fix. It improves recovery but leaves residual bias and can add edge artifacts.8
- Recovery curves are system- and protocol-specific. Measure them on your own scanner, match the VOI method, and report whether correction was applied.37
Conclusion
Quantitative SPECT has earned a place in therapy planning and response assessment, but a number is only as good as the corrections behind it. The partial volume effect is the correction most often underestimated, because it masquerades as a small, cosmetic blur while quietly halving the measured activity of exactly the small lesions clinicians care about most. The remedy is not exotic: measure a recovery curve on your own system with the clinical protocol, apply it consistently with a defined VOI method, validate against an independent object, and always report whether correction was used. Facilities that treat partial volume correction as a standard part of the quantification chain—rather than an afterthought—will produce SUVs and dosimetry that are accurate, reproducible, and defensible.
How DRPS Can Help
Diagnostic Radiation Physics Services helps nuclear medicine and theranostics programs build and validate quantitative SPECT/CT, including system calibration, recovery-coefficient measurement, partial volume correction, and dosimetry workflow review. Our support spans PET/CT and nuclear medicine physics, medical physicist consulting, and accreditation support, aligned with MIRD and EANM guidance and the applicable NRC or Agreement State requirements.
DRPS serves facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware.
Accurate small-object quantification is what turns a pretty picture into a number a physician can act on.
Related Resources
- Quantitative SPECT/CT calibration
- SPECT resolution recovery reconstruction
- SPECT/CT attenuation correction
- Voxel-based dosimetry for radiopharmaceutical therapy
- Red marrow dosimetry for radiopharmaceutical therapy
- PET/CT & nuclear medicine physics services
- Medical physicist consulting
References
- Soret M, Bacharach SL, Buvat I. Partial-volume effect in PET tumor imaging. J Nucl Med. 2007;48(6):932-945. doi:10.2967/jnumed.106.035774. doi.org
- Dewaraja YK, Frey EC, Sgouros G, et al. MIRD pamphlet No. 23: quantitative SPECT for patient-specific 3-dimensional dosimetry in internal radionuclide therapy. J Nucl Med. 2012;53(8):1310-1325. doi:10.2967/jnumed.111.100123. doi.org
- Dickson JC, Armstrong IS, Gabiña PM, et al. EANM practice guideline for quantitative SPECT-CT. Eur J Nucl Med Mol Imaging. 2023;50(4):980-995. doi:10.1007/s00259-022-06028-9. doi.org
- Finocchiaro D, Berenato S, Grassi E, et al. Partial volume effect of SPECT images in PRRT with 177Lu labelled somatostatin analogues: A practical solution. Phys Med. 2019;57:153-159. doi:10.1016/j.ejmp.2018.12.029. doi.org
- 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. J Nucl Med. 2016;57(1):151-162. doi:10.2967/jnumed.115.159012. doi.org
- Ramonaheng K, van Staden JA, du Raan H. The effect of calibration factors and recovery coefficients on 177Lu SPECT activity quantification accuracy: a Monte Carlo study. EJNMMI Phys. 2021;8(1):27. doi:10.1186/s40658-021-00365-8. doi.org
- Morphis M, van Staden JA, du Raan H, Ljungberg M. Evaluation of Iodine-123 and Iodine-131 SPECT activity quantification: a Monte Carlo study. EJNMMI Phys. 2021;8(1):61. doi:10.1186/s40658-021-00407-1. doi.org
- Liu Y, Lu Z, Chen G, Shi K, Mok GSP. Partial volume correction for Lu-177-PSMA SPECT. EJNMMI Phys. 2024;11(1):93. doi:10.1186/s40658-024-00697-1. doi.org
- Jalilifar M, Sadeghi M, Emami-Ardekani A, Geravand K, Geramifar P. Quantifying partial volume effect in SPECT and planar imaging: optimizing region of interest for activity concentration estimation in different sphere sizes. Nucl Med Commun. 2024;45(6):487-498. doi:10.1097/MNM.0000000000001835. doi.org
- Pretorius PH, King MA. Diminishing the impact of the partial volume effect in cardiac SPECT perfusion imaging. Med Phys. 2009;36(1):105-115. doi:10.1118/1.3031110. doi.org
- Mohd Rohani MF, Mat Nawi N, Shamim SE, et al. Maximum standardized uptake value from quantitative bone single-photon emission computed tomography/computed tomography in differentiating metastatic and degenerative joint disease of the spine in prostate cancer patients. Ann Nucl Med. 2020;34(1):39-48. doi:10.1007/s12149-019-01410-4. doi.org
- van Gils CAJ, Beijst C, van Rooij R, de Jong HWAM. Impact of reconstruction parameters on quantitative I-131 SPECT. Phys Med Biol. 2016;61(14):5166-5182. doi:10.1088/0031-9155/61/14/5166. doi.org
Related Articles
-
Quantitative SPECT/CT: Calibration & SUV
Quantitative SPECT/CT turns counts into Bq/mL: system calibration factor, attenuation and scatter correction, recovery coefficients, SUV, and QC for dosimetry.
-
Count Rate and Dead Time in Nuclear Medicine
Count rate performance and dead time set how many events a gamma camera records before losses distort images and quantitation. Here is the physics and QC.
-
Brain Perfusion SPECT: HMPAO and ECD Imaging
Brain perfusion SPECT with Tc-99m HMPAO and ECD maps cerebral blood flow. A physics guide to tracers, acquisition, QC, correction, and quantification.