High-contrast artifact reduction in cone beam computed tomography by using geometric techniques

Peter B. Noël, Jinhui Xu, Kenneth R. Hoffmann, Jason J. Corso, Sebastian Schäfer, Alan M. WalczakProceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2009 · DOI

TL;DR

This paper introduces a novel geometric approach to significantly reduce artifacts in Cone Beam Computed Tomography (CBCT) caused by high-contrast objects like metal implants. By leveraging computational geometry and a dual-space representation, the algorithm efficiently identifies and removes problematic projection rays, leading to clearer reconstructions in real-time.

What Problem Does This Paper Address?

Cone Beam Computed Tomography (CBCT) is a powerful imaging technique offering high-resolution 3D data quickly, making it invaluable for interventions. However, its reconstructions are often plagued by artifacts, particularly from high-contrast objects (e.g., metal implants, dental fillings, surgical clips). These artifacts, stemming from issues like insufficient angular sampling or beam hardening, can obscure crucial details and reduce diagnostic quality. Existing methods for artifact reduction can be computationally intensive or require complex manual intervention. This paper aims to provide an efficient, automated solution to mitigate these artifacts.

Methods

The core of the proposed method lies in a novel application of computational geometry. Instead of directly processing the problematic pixels in the reconstruction, the algorithm first performs a lower-resolution reconstruction to identify the high-contrast objects. Then, it transforms the problem into a "dual space" using point-line duality. In this dual space, the projection rays and object boundaries are represented as points and lines, respectively. A technique called "topological walk/peeling" is then employed to efficiently determine which projection rays intersect the high-contrast objects. This geometric approach models the problem as a visibility problem, allowing for optimal time and space complexity. Once the intersecting rays are identified, they are effectively excluded from the reconstruction process, leading to a cleaner final image. The entire process is designed to be performed in real-time, independent of the projection image size.

Datasets

  • Simulations: A modified Shepp-Logan phantom was used. This phantom included overlapping semi-circles, with one high-contrast object and three smaller low-contrast circles. The data was reprojected into 360 views.
  • Animal Study: Real CBCT data from a rabbit study was utilized. This data consisted of 106 projections with a metal sphere attached to the rabbit's skin, which served as the high-contrast object causing intense artifacts.

Key Insights and Ideas

  • Dual Space Transformation: Representing projection rays and object boundaries in a dual space dramatically simplifies the problem of identifying ray-object intersections. This geometric abstraction is key to the algorithm's efficiency.
  • Topological Walk/Peeling: This computational geometry technique provides an optimal way to navigate the arrangement of lines in the dual space, efficiently identifying the relevant intersections.
  • Real-time Artifact Reduction: By decoupling the artifact identification from the pixel-level reconstruction and leveraging efficient geometric algorithms, the method achieves real-time performance, a significant advantage for clinical applications.
  • Sinogram-Space Correction: Identifying and removing problematic rays in the sinogram (projection) space before reconstruction is more efficient and effective than trying to correct artifacts in the final 3D volume.

Key Results

  • Significant Artifact Reduction: Both simulated and real-world data showed a substantial reduction in artifacts caused by high-contrast objects.
  • Improved Image Quality: Low-contrast objects, often obscured by artifacts, became significantly clearer and more visible in the corrected reconstructions.
  • Real-time Performance: The algorithm was demonstrated to run in real-time, with the animal study correction taking approximately 2 seconds.
  • Accuracy: The identification of intersecting rays was accurate to within 1 pixel.

Limitations and Future Work

  • Initial Reconstruction Dependency: The accuracy of the object identification step relies on an initial, albeit lower-resolution, reconstruction. The paper notes that the number of projections and resolution for this initial step depend on object size and system parameters.
  • Overestimation of Error: The paper mentions that the error in ray identification is always an overestimate, which is a conservative approach but might lead to slightly more data being excluded than strictly necessary.
  • Future Work: The authors suggest that due to the optimal time complexities of their geometric approach, the method's performance will not degrade significantly with larger datasets, making it attractive for future applications. The simplicity of implementation is also highlighted as a benefit for broader adoption.

Summary updated

Abstract

The use of cone beam computed tomography (CBCT) is growing in the clinical arena, due to its ability to provide 3-D information during interventions, its high diagnostic quality (sub-millimeter resolution), and its short scanning times (10 seconds). In many situations, the reconstructions suffer from artifacts from high contrast objects (due mainly to angular sampling by the projections or by beam hardening) which can reduce image quality. In this study, we propose a novel algorithm to reduce these artifacts. In our approach, these objects are identified and then removed in the sinogram space by using computational geometry techniques. In particular, the object is identified in a reconstruction from a few views. Then, the rays (projection lines) intersecting the high contrast objects are identified using the technique of topological walk in a dual space which effectively models the problem as a visibility problem and provides a solution in optimal time and space complexity. As a result, the corrections can be performed in real time, independent of the projection image size. Subsequently, a full reconstruction is performed by leaving out the high contrast objects in the reconstructions. Evaluations were performed using simulations and animal studies. The artifacts are significantly reduced when using our approach. This optimal time and space complexity and relative simple implementation makes our approach attractive for artifact reduction.
BibTeX
@inproceedings{noel2009high,
  author = {Peter B. Noël and Jinhui Xu and Kenneth R. Hoffmann and Jason J. Corso and Sebastian Schäfer and Alan M. Walczak},
  title = {{High-contrast artifact reduction in cone beam computed tomography by using geometric techniques}},
  booktitle = {Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE},
  year = {2009},
  doi = {10.1117/12.811679},
}
Figure