CT Slice Thickness QC and Sensitivity Profiles

In CT, the reconstructed slice thickness is not a physical cut through the patient — it is the full width at half maximum (FWHM) of the slice sensitivity profile (SSP), the scanner's response along the z-axis. Every question about z-axis resolution, partial-volume averaging, image noise, and slice-thickness quality control flows from that one definition. A slice labeled "1.25 mm" is a promise about the shape of a curve, and verifying that promise is a core part of CT acceptance testing and the annual physics survey.¹²

This article explains what the SSP is, how beam collimation and detector geometry shape it, how a physicist measures effective slice width, why thin slices trade noise for resolution, and how slice thickness fits into current accreditation and standards frameworks. DRPS performs this testing as part of its CT physics testing and accreditation support services across Florida, Maryland, Virginia, Washington DC, California, and beyond.

Introduction

When a radiologist selects a 3 mm reconstruction, they are trusting that each transverse image represents a 3 mm-thick slab of anatomy, cleanly bounded above and below. The physical reality is softer. The scanner's sensitivity to a point source does not switch on and off sharply at the slice edges; it rises, plateaus (or peaks), and falls across a finite range of z-positions. That response curve is the slice sensitivity profile, and by international convention its FWHM is the nominal tomographic section thickness.¹

Understanding the SSP matters because it links three things clinicians care about — longitudinal (z-axis) resolution, partial-volume artifact, and image noise — to something a physicist can measure and track. A broadened SSP silently degrades small-structure detectability and multiplanar reformats even when the console still displays the requested slice value. That is exactly why slice width belongs in the quality-control (QC) program, and why the way we measure it deserves careful attention.²³

Topic Explanation

From beam to profile: what the SSP represents

Formally, the SSP is the relative response of a CT system as a function of position along a line perpendicular to the tomographic plane — the z-axis, parallel to the patient's long axis.¹ Imagine a vanishingly thin, high-contrast object (a foil or bead) that you step through the imaged plane one small increment at a time. At each z-position you record how strongly that object contributes to the reconstructed image. Plot contribution versus z, and you have the SSP.

Two features of that curve define the slice:

• FWHM — the width of the profile at half its maximum height — is the reported slice thickness. This is the number on the console.
• FWTM (full width at tenth maximum) describes the tails. Long tails mean signal from anatomy well outside the nominal slice still leaks into the image, degrading contrast for small objects even when the FWHM looks correct.

An idealized SSP is often approximated as a Gaussian or a rounded rectangle. For a Gaussian profile of standard deviation $\sigma$, the FWHM follows directly from the geometry of the curve:

so a slice reported as 2.4 mm corresponds to a Gaussian SSP with $\sigma \approx 1.0$ mm. Real profiles are not perfectly Gaussian — helical interpolation and reconstruction filters distort the shape — which is exactly why we measure the SSP rather than assume it.⁶

Nominal, acquired, and reconstructed thickness

In single-slice CT, the pre-patient z-collimator width largely set the slice thickness. In modern multidetector CT (MDCT), the situation is layered:

• Acquired (detector) collimation — the physical detector-row width used during the scan (for example, $64\times 0.625$ mm or $128\times 0.6$ mm with z-flying focal spot). This sets the finest slice you can reconstruct.
• Reconstructed slice thickness — the thickness you choose at reconstruction, which must be equal to or greater than the detector-row width. The same raw data can produce 0.625 mm, 2.5 mm, or 5 mm images.
• Nominal slice thickness — the value the operator selects and the console indicates, which the SSP FWHM is meant to reproduce.¹

Because reconstructed slice width is a computed quantity derived from the acquired data and the interpolation algorithm, an important modern nuance follows: on today's MDCT scanners, slice thickness rarely drifts on its own the way a mechanical z-collimator once could. We return to what that means for QC below.⁴

Key Technical Principles

Measuring the SSP: the ramp geometry

You cannot lay a ruler on a slice, so physicists exploit geometry. Image a thin, high-contrast object inclined at a known angle to the scan plane — a wire ramp, an angled foil, or a thin plane of material such as those in the standard CT performance and accreditation phantoms.²⁵ Because the object is tilted, a single transverse image intersects it over a range of z, and the object's apparent length in the image plane encodes the slice thickness.

For an object inclined at angle $\theta$ relative to the scan (x–y) plane, a profile length $L$ measured in the image at half maximum corresponds to an effective slice width:

Wire-ramp phantoms commonly use a shallow angle to amplify the signal. For a widely used $\theta =23^{\circ }$ ramp, $\tan 23^{\circ }\approx 0.42$, so a measured FWHM length of, say, 12 mm in the image plane corresponds to:

Automated tools now segment the ramp or stair object, correct for its angle (for example via a Hough transform), extract the intensity profile, and compute the FWHM directly — one study reported automated slice-thickness results agreeing with nominal values to within 1.0 mm and with manual caliper measurement to within about 12% across filters and off-center positions.⁸ The alternative, more direct method steps a thin bead or foil along z in small increments and samples the SSP point by point; it is the reference approach but is slower and demands precise couch indexing.

The dose profile is wider than the sensitivity profile

A subtle but important distinction: the radiation dose profile along z is generally wider than the sensitivity profile. The x-ray beam is collimated to the detector, but geometric penumbra, the finite focal-spot size, and the anode heel effect spread the delivered radiation beyond the active detector rows.⁷ To ensure the outermost detector channels see a flat, fully-penumbra-free beam, manufacturers deliberately overbeam — illuminating slightly more than the active detectors — which is one reason narrow-collimation scans are dose-inefficient. Analytical dose-profile models that incorporate anode tilt and the heel effect show measurable asymmetry in the penumbra that simple rectangular-beam assumptions miss.⁷ The practical message: slice-width QC characterizes image sensitivity, while geometric efficiency and overbeaming are separate dose considerations addressed elsewhere in the physics survey.

Helical broadening and z-sampling

In helical (spiral) acquisition, the table moves during rotation, so no single rotation directly yields a planar slice; the image is interpolated from data acquired over a helix. This interpolation broadens the SSP relative to the axial case, and the broadening grows with pitch. Two engineering advances tamed it:

• 180° linear interpolation and later cone-beam reconstruction, which keep the effective SSP close to nominal across a range of pitches.
• The z-flying focal spot, which periodically deflects the focal spot in the z-direction to double the longitudinal sampling density. On a 64-slice scanner using $32\times 0.6$ mm collimation, this technique acquires 64 overlapping 0.6 mm samples per rotation, with measured spiral SSP FWHM differing from the nominal values (0.6, 0.75, 1.0, 1.5, 2.0 mm) by less than 0.15 mm, and the thinnest 0.6 mm slice measuring 0.66–0.68 mm FWHM at isocenter.⁶ That is the quantitative reason modern MDCT holds slice width so tightly — and why slice thickness has become a less volatile QC parameter.

Slice thickness, noise, and z-resolution: the trade-off

Thinner slices are not free. For a fixed technique (kVp, mAs, pitch), image noise standard deviation ${\sigma }_{\text{noise}}$ scales inversely with the square root of the slice thickness $T$:

because a thinner voxel integrates fewer detected photons. Equivalently, the product of image noise and $\sqrt{T}$ is approximately constant for a given dose — a relationship confirmed experimentally across pitch on z-flying-focal-spot systems.⁶ Halving the slice thickness therefore multiplies noise by $\sqrt{2}\approx 1.41$; going from 5 mm to 0.625 mm multiplies noise by $\sqrt{5/0.625}=\sqrt{8}\approx 2.83$. The table below makes the trade explicit.

Measured-FWHM values are representative of a modern z-flying-focal-spot MDCT at isocenter⁶ and will vary with scanner, reconstruction kernel, pitch, and off-center position; they are illustrative, not a substitute for a scanner-specific measurement.

Partial-volume averaging: the clinical cost of a thick slice

When a structure does not span the full slice thickness, the reconstructed CT number is a volume-weighted average of the structure and whatever else shares the voxel — the partial volume effect. If a structure of CT number $H_{\text{obj}}$ occupies a fraction $f$ of the slice thickness and background $H_{\text{bg}}$ fills the rest, the measured CT number is approximately:

Consider a 2 mm calcified focus ($H_{\text{obj}}=400$ HU) imaged in a 5 mm slice against soft-tissue background ($H_{\text{bg}}=40$ HU). The fraction is $f=2/5=0.4$, so:

The focus reads 184 HU instead of 400 HU — over-averaged, lower-contrast, and easy to miss. Reconstruct the same raw data at 1.25 mm and $f\to 1$ across the central slice, recovering the true CT number and sharpening the lesion boundary. This is why thin slices and z-resolution matter for small pulmonary nodules, subtle calcifications, and small-vessel CTA, and it connects directly to CT image quality, MTF, and low-contrast detectability and CT number/HU calibration.

Clinical Impact

A well-controlled SSP is invisible when everything works and expensive when it does not. The three failure modes that a slice-width program guards against are:

• Silent z-axis blur. A broadened SSP (long tails, elevated FWTM) degrades multiplanar reformats and 3D renderings that radiologists and surgeons increasingly rely on, even though axial images can look acceptable. Reformats are only as good as the z-axis resolution feeding them.
• Under-recovered small lesions. Partial-volume averaging in thick slices suppresses the contrast of sub-slice structures — a recurring theme in lung-nodule and coronary-calcium work — so protocol slice selection is a diagnostic decision, not just a data-management one.
• Noise surprises. When a site pushes to ultra-thin reconstructions for reformats without adjusting technique, images can become too noisy to read. Knowing the $1/\sqrt{T}$ relationship lets the physicist and technologist choose a reconstruction thickness that balances z-resolution against noise — often reconstructing thin for reformats and thicker for primary axial review from the same acquisition.⁶

These considerations feed directly into CT protocol optimization: the "right" slice thickness is the one that answers the clinical question at acceptable noise and dose, which is rarely the thinnest option available.

Practical Optimization Tips

1. Measure the SSP at acceptance, not just the FWHM number. Capture the full profile shape so you can watch FWTM/tails, not only the headline slice width. Baseline every scanner and reconstruction mode you will use clinically.³
2. Test at more than one nominal thickness. Verify a thin, a mid, and a thick reconstruction. A scanner can hold 5 mm perfectly and still broaden badly at 0.625 mm.
3. Check off-center behavior. SSP and z-resolution degrade away from isocenter. Automated methods have confirmed stable results within a few centimeters of isocenter, but larger offsets and cone-beam effects at wide collimation deserve a look.⁶⁸
4. Match reconstruction increment to slice thickness for reformats. Overlapping reconstruction (increment < slice thickness, e.g., 50% overlap) improves reformat quality without a new scan; it does not change the SSP FWHM but it improves z-sampling of the volume.
5. Reconstruct thin for reformats, thicker for review. Leverage the single acquisition: keep a thin series for 3D/MPR and a thicker, lower-noise series for primary interpretation.
6. Re-test after major service or software upgrades. Reconstruction-algorithm changes (including iterative and deep-learning reconstruction) can alter effective z-response; see iterative and deep-learning reconstruction. Verify slice width and noise behavior after such changes.
7. Use a qualified phantom correctly. Ramp angle, phantom leveling, and window settings all bias a ramp-based FWHM. Follow the phantom's instructions precisely — small setup errors are the leading cause of erroneous slice-width results in accreditation submissions.⁵

Regulatory Considerations

Slice-thickness accuracy sits at the intersection of international performance standards, accreditation requirements, and the annual medical-physics survey. The governing references are:

• IEC 60601-2-44 (Edition 3.2, the 2009 third edition consolidated with its 2012 and 2016 amendments) defines the sensitivity profile and the nominal tomographic section thickness and sets the essential-performance framework for CT x-ray equipment. It is the anchor standard against which manufacturer slice-width specifications are written.¹
• IEC 61223-3-5 provides acceptance-test methodology for CT, including image-quality parameters that a physicist verifies at installation.¹⁰
• The ACR CT Quality Control Manual structures the routine QC program and the annual physicist tests. Notably, the current manual treats slice thickness as no longer a required annual test for modern scanners, on the reasoning that reconstructed slice width is a computed quantity that is not an independent failure mode; the manual notes it can still be valuable for the few single-slice scanners still in use.³⁴ This is an important, current nuance: it does not mean z-axis performance is unimportant — it means routine drift is unlikely on MDCT, so the test's yield at annual cadence is low. Physicists still verify slice width at acceptance and whenever a reconstruction change warrants it.
• The ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of CT describes the qualified medical physicist's role in the CT performance evaluation, within which z-axis and image-quality performance are assessed.⁹
• AAPM Report No. 39 remains the foundational description of CT scanner specification and acceptance testing, including slice-width measurement, and informs how these tests are performed today.²

Because CT scanners are x-ray-producing devices, they are regulated at the federal level by the FDA and, for their use and registration, by state radiation-control programs — for example, Florida's program under Chapter 64E-5, F.A.C. — rather than by the NRC, which governs radioactive material. A qualified medical physicist's CT performance evaluation, including baseline z-axis characterization, supports both accreditation and state compliance. For the broader picture, see ACR CT accreditation physics requirements.

Frequently Asked Questions (FAQs)

What is the slice sensitivity profile in CT?

The slice sensitivity profile (SSP) is the relative response of a CT system as a function of position along the z-axis, perpendicular to the image plane. The nominal slice thickness is defined as the full width at half maximum (FWHM) of the SSP. A wide SSP means the reconstructed image blends signal from a thicker slab of anatomy.

How is CT slice thickness actually measured?

Slice thickness is measured from the SSP rather than with a ruler. A physicist images a thin high-contrast object angled relative to the scan plane — a wire ramp, an inclined bar, or a thin bead stepped along z. The FWHM of the resulting profile, corrected for the ramp geometry ($T=L\tan \theta$), gives the effective slice width in millimeters.

Why do thin CT slices look noisier?

Image noise scales approximately as one over the square root of slice thickness for a fixed radiation dose. Cutting the slice from 5 mm to 1.25 mm doubles the noise standard deviation because each thinner voxel collects fewer photons. Thin slices improve z-axis resolution and reduce partial-volume averaging but cost noise unless dose is increased.

Is slice thickness still a required CT accreditation test?

In the current ACR CT Quality Control Manual, slice thickness is no longer a required annual test for modern multidetector scanners because reconstructed slice width is a computed quantity that rarely fails independently. It remains valuable at acceptance testing, for single-slice scanners, and whenever z-axis performance or a reconstruction change is in question.

What is the partial volume effect and how does slice thickness affect it?

The partial volume effect occurs when a structure only partly fills the slice thickness. The reconstructed CT number becomes a volume-weighted average of the structure and its surroundings, so a small high-contrast object measures a falsely low CT number and blurs into adjacent voxels. Thinner slices reduce partial-volume averaging and improve small-lesion detectability.

What slice-thickness tolerance should a CT scanner meet?

Acceptance criteria are typically referenced to the manufacturer specification and international standards such as IEC 60601-2-44, with measured slice width expected to agree with the nominal value within a small tolerance that widens for the thinnest slices. Modern multidetector helical scanners with z-flying focal-spot sampling routinely hold measured SSP FWHM within a few tenths of a millimeter of nominal.

Who should perform CT slice-thickness and SSP testing?

SSP and slice-width measurements are performed or supervised by a qualified medical physicist as part of acceptance testing and the annual CT equipment performance evaluation. DRPS provides this testing as part of its CT physics and accreditation support services.

Key Takeaways

• Slice thickness is the FWHM of the slice sensitivity profile, not a physical cut — every z-axis performance question follows from the SSP.¹
• Measurement uses geometry, not a ruler: an angled ramp or a stepped bead yields a profile whose FWHM, corrected by $T=L\tan \theta$, is the effective slice width.⁸
• Thin slices trade noise for resolution: noise scales as $1/\sqrt{T}$, so halving thickness raises noise by ~41%.⁶
• Partial-volume averaging suppresses sub-slice lesions; thinner reconstruction recovers true CT number and small-object contrast.
• Modern MDCT holds slice width tightly thanks to helical interpolation and z-flying-focal-spot sampling (FWHM within ~0.15 mm of nominal), which is why the ACR manual no longer requires it as an annual test.⁴⁶
• Verify at acceptance and after reconstruction changes; baseline multiple thicknesses and off-center positions.

Conclusion

The slice sensitivity profile turns an abstract label — "1.25 mm" — into a measurable, trackable performance characteristic. Its FWHM sets slice thickness; its tails govern how cleanly the slice is bounded; and its interaction with detector geometry, helical interpolation, and reconstruction determines the z-axis resolution, partial-volume behavior, and noise that a radiologist ultimately sees. Modern scanners hold slice width so well that routine annual testing has been de-emphasized, but that stability is precisely a QC achievement, not a reason to stop understanding the physics. A physicist who characterizes the SSP at acceptance, re-checks it after reconstruction changes, and helps a site choose slice thickness deliberately protects both image quality and dose efficiency.

How DRPS Can Help

Diagnostic Radiation Physics Services performs CT acceptance testing and annual equipment performance evaluations, including slice sensitivity profile characterization, z-axis resolution, image noise, CT number accuracy, and dose measurement, and helps sites translate the results into practical protocol and reconstruction decisions. Our physicists support CT physics testing, accreditation support, and medical physicist consulting across our service areas, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware. See our service locations or contact us to schedule a CT survey.

Related Resources

• CT Image Quality: MTF and Low-Contrast Detectability
• CT Number (HU) Calibration QC
• CT Protocol Optimization
• CT Iterative and Deep-Learning Reconstruction
• ACR CT Accreditation Physics Requirements
• CT Physics Testing services
• Accreditation Support services

References

1. International Electrotechnical Commission. IEC 60601-2-44:2009+AMD1:2012+AMD2:2016 (Edition 3.2): Medical electrical equipment — Part 2-44: Particular requirements for the basic safety and essential performance of X-ray equipment for computed tomography. iec.ch
2. American Association of Physicists in Medicine. AAPM Report No. 39: Specification and Acceptance Testing of Computed Tomography Scanners. 1993. aapm.org
3. American College of Radiology. ACR Computed Tomography Quality Control Manual. acr.org
4. Butler PF. ACR frequently asked questions: updates to the ACR CT Quality Control Manual. AAPM Newsletter. 2017;Nov/Dec. aapm.org
5. McCollough CH, Bruesewitz MR, McNitt-Gray MF, et al. The phantom portion of the American College of Radiology (ACR) computed tomography (CT) accreditation program: practical tips, artifact examples, and pitfalls to avoid. Med Phys. 2004;31(9):2423-2442. doi:10.1118/1.1769632. PubMed
6. Flohr TG, Stierstorfer K, Ulzheimer S, Bruder H, Primak AN, McCollough CH. Image reconstruction and image quality evaluation for a 64-slice CT scanner with z-flying focal spot. Med Phys. 2005;32(8):2536-2547. doi:10.1118/1.1949787. PubMed
7. Dixon RL, Munley MT, Bayram E. An improved analytical model for CT dose simulation with a new look at the theory of CT dose. Med Phys. 2005;32(12):3712-3728. doi:10.1118/1.2122507. PubMed
8. Lasiyah N, Anam C, Hidayanto E, Dougherty G. Automated procedure for slice thickness verification of computed tomography images: variations of slice thickness, position from iso-center, and reconstruction filter. J Appl Clin Med Phys. 2021;22(7):313-321. doi:10.1002/acm2.13317. PubMed
9. American College of Radiology. ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of Computed Tomography (CT) Equipment. acr.org
10. International Electrotechnical Commission. IEC 61223-3-5: Evaluation and routine testing in medical imaging departments — Part 3-5: Acceptance tests — Imaging performance of computed tomography X-ray equipment. iec.ch