CT-Based Attenuation Correction in PET/CT

CT-based attenuation correction is the step that turns PET from a qualitative picture into a quantitative measurement. It converts the CT image into a map of how strongly tissue attenuates 511 keV annihilation photons, then uses that map to recover the true activity distribution — the foundation of an accurate standardized uptake value (SUV). ^{1, 2}

Attenuation is the single largest physical correction in PET: without it, deep structures look falsely cold and superficial tissue looks falsely hot. Modern PET/CT derives the correction from the CT scan, which is fast, low-noise, and inherently co-registered. But borrowing the CT to estimate attenuation at a different photon energy introduces its own pitfalls. This guide explains how the conversion works, the artifacts it can create, and the quality control that keeps quantification trustworthy. DRPS provides this analysis through its PET/CT and nuclear medicine physics services. ^{1, 2, 3}

Introduction

Roughly speaking, only a fraction of the annihilation photon pairs created inside the body actually escape to be detected, and that fraction depends strongly on how much tissue they must traverse. A pair of 511 keV photons emitted near the center of the torso may have well under a tenth the chance of both escaping compared with a pair emitted near the skin. If uncorrected, this makes the reconstructed image non-quantitative and visually misleading — a phenomenon every nuclear medicine physicist learns to recognize. ¹

Early dedicated PET scanners measured attenuation directly with rotating radioactive transmission sources (germanium-68 rods or cesium-137 points), acquiring a "transmission scan" that was noisy and slow. The introduction of the combined PET/CT scanner replaced that transmission source with the CT subsystem, which maps tissue attenuation in seconds with far less noise and perfect inherent registration. ¹ This was a major reason PET/CT displaced standalone PET so rapidly.

The catch is that CT measures attenuation using a polyenergetic X-ray beam with a mean energy on the order of 70 keV, while PET photons are monoenergetic at 511 keV. Attenuation is energy-dependent and tissue-dependent, so the CT image must be transformed, not simply reused. How well that transformation is done — and how its failure modes are recognized — determines whether SUVs can be trusted. This topic sits alongside PET SUV quantification and EARL SUV harmonization as a core quantitative-PET fundamental.

Topic Explanation

Attenuation along a line of response

PET detects two photons in coincidence along a line of response (LOR). The probability that both photons escape the body depends on the total attenuation along the entire LOR, not on where along the line the annihilation occurred. For a LOR passing through tissue with linear attenuation coefficient $\mu (x)$ at 511 keV, the survival probability is: ¹

The attenuation correction factor (ACF) applied to that LOR is the reciprocal:

Because this integral is taken over the whole path, PET attenuation correction has a convenient property: the correction can be computed from a transmission measurement (or CT) independent of where the activity actually is. The job of CT-based attenuation correction is therefore to supply an accurate map of $\mu (x)$ at 511 keV everywhere in the body. ¹

Why a CT image is not a 511 keV map

A CT image is expressed in Hounsfield units (HU), which are referenced to water at the CT beam's effective energy:

These attenuation coefficients are at CT energies (tens of keV), where the photoelectric effect contributes significantly and depends steeply on atomic number. At 511 keV, attenuation is dominated by Compton scatter, which depends mainly on electron density. As a result, the ratio of bone-to-soft-tissue attenuation is very different at 70 keV than at 511 keV. A single global scaling factor cannot convert the whole HU range correctly — bone and soft tissue need different treatment. ^{1, 2}

Key Technical Principles

The bilinear conversion

The standard solution, used by all major vendors, is a bilinear (two-segment) scaling of HU to the 511 keV linear attenuation coefficient ${\mu }_{511}$. ^{1, 2} The pivot is at water (HU = 0). Taking the linear attenuation coefficient of water at 511 keV as approximately ${\mu }_{w,511}\approx 0.096{\text{cm}}^{-1}$:

For HU at or below 0, the body is a mixture of air and soft tissue, whose attenuation scales nearly the same way at CT and PET energies, so a simple water-referenced scaling works. For HU above 0, the contribution of bone (higher atomic number, more photoelectric absorption at CT energies) means a different, shallower slope $k$ is required, fitted so that cortical bone maps to its correct 511 keV coefficient. The exact breakpoint slopes depend on the CT tube voltage (kVp), which changes the beam spectrum — so vendors provide kVp-specific conversion curves. ^{1, 2}

Worked conversion example

Apply the soft-tissue segment to a few representative tissues, using ${\mu }_{w,511}=0.096{\text{cm}}^{-1}$:

• Lung (HU ≈ −700): ${\mu }_{511}=0.096(1-0.700)=0.096\times 0.300\approx 0.029{\text{cm}}^{-1}$
• Water / blood (HU = 0): ${\mu }_{511}=0.096(1+0)=0.096{\text{cm}}^{-1}$
• Soft tissue (HU ≈ +40): ${\mu }_{511}=0.096(1+0.040)\approx 0.100{\text{cm}}^{-1}$

Cortical bone (HU ≈ +1000) falls on the upper segment and maps to roughly $0.17{\text{cm}}^{-1}$ at 511 keV, illustrating why the second slope is needed: a naive linear extension of the soft-tissue segment would overestimate bone attenuation. ^{1, 2} These attenuation coefficients are then integrated along each LOR to build the ACF map.

The artifact table

Because the attenuation map is derived from CT, anything that makes the CT non-representative of the tissue's true 511 keV attenuation — or that misaligns the CT with the PET — produces an artifact. The table summarizes the major ones. ^{2, 4, 5}

A practical rule follows from this table: always read the attenuation-corrected and non-attenuation-corrected (NAC) PET images together. A true focus of uptake is present on both; an attenuation-correction artifact often appears only on the corrected image. This single habit prevents most artifact-driven misreads. ^{4, 5}

Respiratory mismatch in detail

The most common clinically significant artifact is respiratory mismatch at the lung–liver interface. The CT may be acquired at one point in the breathing cycle while PET averages over many cycles. If the diaphragm position differs, the CT places lung attenuation where PET sees liver, or vice versa, producing a characteristic crescent of artifactually reduced activity above the diaphragm and potentially mispositioning a lung-base or hepatic-dome lesion. Using a quiet mid-expiration breathing protocol for the CT used in attenuation correction substantially reduces this mismatch. ^{2, 4}

Clinical Impact

Attenuation correction is not a back-room technical detail — it directly shapes diagnosis, staging, and treatment response assessment. Three consequences are worth emphasizing.

SUV reliability. SUV is the workhorse semi-quantitative metric in oncologic PET, used to characterize lesions and to track response across serial scans. Because SUV is computed from attenuation-corrected activity, a biased attenuation map biases SUV. Phantom and patient studies have documented that CT-based attenuation correction tracks reference methods closely in uniform tissue but can deviate by tens of percent in the presence of contrast or metal — large enough to change a clinical interpretation if unrecognized. ¹ Consistency of attenuation correction is also central to multi-center SUV harmonization efforts such as EARL. ³

Lesion detection and localization. Respiratory mismatch and truncation can both create or hide apparent lesions near the diaphragm and body margins. Recognizing these as artifacts — rather than disease — avoids unnecessary biopsies and follow-up imaging. ^{4, 5}

Quantitative therapy assessment. In theranostics and response-adapted therapy, the same patient is imaged repeatedly and small SUV changes carry weight. Stable, well-controlled attenuation correction is what makes those comparisons meaningful, and it connects directly to PET/CT NEMA NU-2 performance testing and routine SPECT/CT and PET/CT quality control.

Practical Optimization Tips

Protect quantitative accuracy

• Always compare AC and NAC images. It is the fastest, cheapest artifact check available and should be routine for every read. ^{4, 5}
• Use a quiet-breathing CT protocol for the attenuation-correction CT to minimize diaphragm mismatch. ^{2, 4}
• Position arms up for body PET/CT when clinically feasible to avoid truncation at the CT field-of-view edge. ⁴
• Be deliberate about contrast. If contrast-enhanced CT is used for attenuation correction, understand the SUV bias it can introduce, or acquire a separate low-dose CT for attenuation correction. ²
• Confirm the kVp-appropriate conversion is selected; the bilinear curve depends on tube voltage. ^{1, 2}
• Verify PET/CT registration at installation, after service, and routinely; misregistration corrupts the attenuation map. ³

Quality control that anchors the correction

• Uniform cylinder phantom: a properly attenuation-corrected image of a uniform cylinder should be flat; cupping or doming reveals attenuation or scatter-correction problems. ³
• Calibration (well counter / SUV calibration): confirm the scanner converts counts to activity concentration correctly, so SUV is anchored to a known standard. ³
• NEMA NU 2 performance methodology: the consensus framework for measuring PET performance, including the count-rate and image-quality tests that underlie quantitative reliability. ⁶
• Accreditation phantom imaging: ACR–AAPM PET/CT accreditation includes phantom-based image-quality and SUV-accuracy checks reviewed by a qualified medical physicist. ⁷

For the broader QC context, see our guides to dose calibrator quality control and PET uptake time standardization.

Regulatory Considerations

PET/CT sits at the intersection of byproduct-material regulation and X-ray machine regulation, and attenuation-correction quality is part of the quantitative-accuracy expectations that accreditation and licensing impose. The positron-emitting radiopharmaceuticals (such as F-18 FDG) are byproduct material regulated under federal or Agreement-State rules; the CT subsystem is regulated as a radiation-producing machine by the FDA and state authorities. ^{8, 9}

• 10 CFR Part 35 (Medical Use of Byproduct Material) governs authorized use, dosage determination, and the radiation safety program for the PET radiopharmaceuticals, with occupational and public dose limits set by 10 CFR Part 20. Agreement States administer equivalent programs. ^{8, 9}
• ACR–AAPM PET/CT Accreditation and equivalent programs require periodic medical-physicist evaluation that includes SUV accuracy, uniformity, and registration — all dependent on sound attenuation correction. ⁷
• SNMMI and EANM procedure standards define the imaging and quantification practices (including attenuation correction and SUV harmonization) expected for clinical and trial-grade PET/CT. ^{3, 5}

Of the jurisdictions DRPS serves, Florida, Maryland, Virginia, California, Nevada, Pennsylvania, New York, and New Jersey are NRC Agreement States, while Washington DC and Delaware are regulated directly by the NRC for byproduct material. Facilities should confirm which authority issues their license and which accreditation body applies. DRPS provides PET/CT and nuclear medicine physics, CT physics testing, and accreditation support across these locations. ^{7, 8}

Frequently Asked Questions (FAQs)

What is attenuation correction in PET/CT?

Attenuation correction compensates for the 511 keV annihilation photons that are absorbed or scattered before leaving the body. In PET/CT, the CT scan is converted into a map of attenuation coefficients at 511 keV, which is used to recover the true activity distribution so PET images and SUVs are quantitatively accurate.

Why is CT used for PET attenuation correction?

CT provides a low-noise, high-resolution map of tissue attenuation in seconds, far faster than the older radioactive transmission sources such as Ge-68 rods or Cs-137 points. Using CT shortens the exam, lowers transmission-scan noise, and is inherently co-registered with the PET data.

How are CT Hounsfield units converted to 511 keV attenuation coefficients?

A bilinear (two-segment) scaling function maps CT numbers to 511 keV linear attenuation coefficients. One segment covers air-to-water tissue (HU below 0) and a second covers water-to-bone (HU above 0), because the energy dependence of attenuation differs for soft tissue versus bone and the CT beam is polyenergetic while PET is monoenergetic.

What artifacts can attenuation correction introduce?

CT-based attenuation correction can introduce artifacts from metal implants, iodinated or oral contrast, respiratory mismatch between the CT and PET, patient motion, and truncation when the patient extends beyond the CT field of view. These can create false uptake or distort SUV near the affected region.

Why should non-attenuation-corrected images also be reviewed?

Comparing the attenuation-corrected and non-corrected PET images helps distinguish true uptake from an attenuation-correction artifact. If a focus appears only on the corrected image and not the non-corrected image, it may be an artifact from metal, contrast, or misregistration rather than real tracer uptake.

Does attenuation correction affect SUV?

Yes. The standardized uptake value depends directly on accurate attenuation correction. Errors in the attenuation map — from contrast, metal, or respiratory mismatch — can bias SUV upward or downward, which is why attenuation-correction quality control is part of any quantitative PET program.

How is attenuation correction quality verified?

Routine PET/CT quality control includes a uniform cylinder phantom to confirm corrected images are flat and quantitatively accurate, calibration against a known activity, PET/CT registration checks, and adherence to NEMA NU 2 performance methodology and accreditation requirements.

Key Takeaways

• Attenuation correction is the largest physical correction in PET and is essential for quantitative, visually accurate images. ¹
• CT replaced radioactive transmission sources because it is fast, low-noise, and inherently co-registered. ¹
• A bilinear HU-to-μ conversion is required because soft tissue and bone scale differently from CT energies to 511 keV. ^{1, 2}
• Derived-from-CT means vulnerable to CT problems: metal, contrast, respiratory mismatch, motion, and truncation all create artifacts. ^{4, 5}
• Read AC and NAC images together to separate true uptake from correction artifacts. ^{4, 5}
• QC anchors quantification: uniform phantom, calibration, registration, NEMA NU 2, and accreditation testing keep SUV trustworthy. ^{3, 6, 7}

Conclusion

CT-based attenuation correction is what makes PET quantitative, and it is also one of the most common sources of artifact in PET/CT interpretation. The physics is elegant: convert the CT image to a 511 keV attenuation map with a bilinear function, integrate along each line of response, and recover the true activity. The failure modes are equally predictable: when the CT does not represent the tissue's real 511 keV attenuation — because of metal, contrast, motion, breathing, or truncation — the correction propagates that error into the PET image and its SUVs. ^{1, 2, 4}

A strong quantitative PET program treats attenuation correction as something to be actively managed, not assumed: kVp-appropriate conversion curves, quiet-breathing CT protocols, routine AC-versus-NAC comparison, verified registration, and phantom-based QC anchored to NEMA NU 2 and accreditation requirements. Done well, attenuation correction is invisible — which is exactly the point. ^{3, 6}

How DRPS Can Help

Diagnostic Radiation Physics Services supports PET/CT and nuclear medicine facilities with PET/CT and nuclear medicine physics testing, SUV calibration and harmonization review, attenuation-correction and registration QC, NEMA NU 2 performance evaluation, ACR PET/CT accreditation support, and medical physics consulting by board-certified medical physicists.

DRPS serves facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware. Trustworthy SUVs start with a trustworthy attenuation map — we help you prove yours is.

Related Resources

• PET SUV quantification
• EARL PET SUV harmonization
• PET/CT NEMA NU-2 performance testing
• SPECT/CT quality control
• PET uptake time standardization
• Dose calibrator quality control
• PET/CT and nuclear medicine physics services
• Medical physicist consulting

References

1. Beyer T, Townsend DW, Brun T, et al. A combined PET/CT scanner for clinical oncology. J Nucl Med. 2000;41(8):1369-1379. PubMed
2. Visvikis D, Costa DC, Croasdale I, et al. CT-based attenuation correction in the calculation of semi-quantitative indices of [18F]FDG uptake in PET. Eur J Nucl Med Mol Imaging. 2003;30(3):344-353. doi:10.1007/s00259-002-1070-4. PubMed
3. Boellaard R, Delgado-Bolton R, Oyen WJG, et al. FDG PET/CT: EANM procedure guidelines for tumour imaging: version 2.0. Eur J Nucl Med Mol Imaging. 2015;42(2):328-354. doi:10.1007/s00259-014-2961-x. PubMed
4. Shammas A, Lim R, Charron M. Pediatric FDG PET/CT: physiologic uptake, normal variants, and benign conditions. RadioGraphics. 2009;29(5):1467-1486. doi:10.1148/rg.295085247. PubMed
5. Xia T, Alessio AM, Kinahan PE. Dual energy CT for attenuation correction with PET/CT. Med Phys. 2014;41(1):012501. doi:10.1118/1.4828838. PubMed
6. National Electrical Manufacturers Association. NEMA NU 2: Performance Measurements of Positron Emission Tomographs (PET). Rosslyn, VA: NEMA. nema.org
7. American College of Radiology. PET/CT Accreditation Program Requirements. Reston, VA: ACR. acr.org
8. U.S. Nuclear Regulatory Commission. 10 CFR Part 35: Medical Use of Byproduct Material. ecfr.gov
9. U.S. Nuclear Regulatory Commission. 10 CFR Part 20: Standards for Protection Against Radiation. ecfr.gov