Visualising Parker Weighting in Short-Scan Cone-Beam Micro-CT: A Practical Reference

Abstract: Short-scan FDK reconstruction is widely used in preclinical cone-beam micro-CT because it reduces scan time and radiation dose, and because the large volume sizes typical of micro-CT make iterative methods impractical for routine use. Short scans, however, introduce non-uniform data redundancy that must be corrected by Parker weighting to avoid directional shading artefacts. This note provides a visual and quantitative summary of Parker weighting as implemented for the eXplore CT 120 scanner. We illustrate the weight maps in the detector and sinogram domains, demonstrate the shading artefacts that arise without correction on both an image quality phantom and an in vivo mouse lung, and show via MTF, NPS, and detectability analysis that Parker weighting corrects HU inaccuracies without degrading image quality. No new methods are introduced; the aim is to serve as a concise practical reference for groups implementing or evaluating short-scan FDK pipelines.

• The paper demonstrates that Parker weighting effectively eliminates directional shading artifacts in short-scan micro-CT reconstructions.
• The methodology utilizes a piecewise Parker weight function with sin² ramps to correct non-uniform data redundancy and ensure quantitative accuracy.
• Quantitative evaluations confirm that Parker weighting preserves spatial resolution and low-contrast detectability, reinforcing its practical value in preclinical imaging.

Parker Weighting in Short-Scan Cone-Beam Micro-CT: Technical Summary and Implications

Motivation and Context

Short-scan Feldkamp-Davis-Kress (FDK) reconstruction is the standard in preclinical cone-beam micro-CT due to its computational efficiency and dose reduction benefits. Large volume sizes characteristic of micro-CT render iterative reconstruction schemes nonviable for routine application. Short-scan acquisition, defined by limited angular coverage ($\pi + 2\gamma_m$), introduces non-uniform data redundancy along the projection arc, leading to significant directional shading artefacts if not adequately corrected. Smooth Parker weighting corrects for these artefacts by adjusting the contribution of doubly-sampled rays, ensuring uniformity and quantitative accuracy in reconstructed attenuation values.

Technical Implementation and Visualisation

The paper presents a detailed visualisation of Parker weighting for the eXplore CT 120 geometry (193º scan arc, $Y_m = 6.5^\circ$), providing graphical representations of weight maps in both detector and sinogram domains. The Parker weight function ($w(\beta, \gamma)$) is derived as a piecewise formulation with sin$^2$ ramps at arc boundaries, guaranteeing smooth transition from zero to unity for peripheral rays, while central rays are fully weighted. The mapping ensures conjugate rays ($$w(\beta, \gamma) + w(\beta + \pi + 2\gamma, -\gamma) = 1$$) have constant total weight, directly addressing inhomogeneous redundancy.

For cone-beam geometry, the standard Parker fan-beam weighting is applied row-wise, relying on the FDK approximation where the cone angle is not explicitly incorporated. The implementation utilised the following pipeline:

1. Preprocessing (dark/flood correction, log transform)
2. Cosine weighting
3. Parker weighting
4. Ramp filtering
5. Voxel-driven backprojection
6. HU calibration

Quantitative Evaluation and Strong Numerical Results

Reconstruction experiments using both a reference image quality phantom and in vivo mouse lung data demonstrate the practical impact of Parker weighting. Omitting Parker weighting produces prominent directional shading artefacts, visible as intensity inhomogeneities and systematic HU errors concentrated at volume boundaries or in regions of complex attenuation. Residual maps reveal spatial patterns of artefactual error—ring-like for the phantom, directional for the mouse lung—which are object-dependent.

Quantitative image quality metrics show no degradation attributable to Parker weighting:

• Modulation Transfer Function (MTF10): Nearly identical values (1.56 lp/mm without weighting vs 1.60 lp/mm with weighting), affirming spatial resolution preservation.
• Noise Power Spectrum (NPS): Slight reduction in apparent noise power in unweighted reconstructions, attributed to deterministic shading rather than true noise suppression.
• Non-Prewhitening Detectability Index ($d'$): Task-based detectability was equivalent across conditions, confirming no impact on low-contrast discrimination for disc signals (at $\Delta$HU = 100, 500).

These results establish that Parker weighting yields quantitatively correct HU maps without penalty to resolution or detectability—contradicting any assumption that weighting may compromise image quality metrics.

Practical and Theoretical Implications

Parker weighting is pivotal for quantitative accuracy in short-scan micro-CT, particularly in domains such as tissue classification, bone density measurement, and longitudinal studies where systematic errors propagate downstream. The object-dependent expression of artefacts necessitates weighting correction in all applications with limited angular coverage.

The Parker scheme is analytically robust and practically straightforward for implementation in standard FDK pipelines. Its extension to three-dimensional cone-beam systems is via the row-wise application; for wider cone angles or advanced acquisition geometries, further generalisation may be required. The study reinforces the necessity for explicit correction even in cases where raw images may appear superficially uniform.

Open-source code availability for the eXplore CT 120 pipeline promotes reproducibility and provides a practical reference for the broader preclinical imaging community.

Speculation on Future Directions

Future work may focus on adaptive weighting for complex objects with varying attenuation, or on systematic analysis of weighting functions in high cone-angle geometries and advanced partial-scan configurations. Continued development of quantitative metrics for pipeline evaluation will enhance comparability and reliability across platforms. Integration with iterative schemes may be revisited as hardware advancements render them less prohibitive.

Conclusion

The study conclusively demonstrates that Parker weighting is crucial for eliminating directional shading artefacts and achieving quantitatively reliable HU values in short-scan cone-beam micro-CT reconstructions. Application of weighting does not impair spatial resolution or detectability, confirming its necessity and practical feasibility. The provided pipeline and visualisation serve as a comprehensive reference for implementation and evaluation of short-scan FDK protocols in micro-CT systems (2604.23390).