Super-sparsely view-sampled cone-beam CT by incorporating prior data.

Area of Science:

• Medical imaging
• Image reconstruction
• Computational imaging

Background:

• Computed tomography (CT) is essential in medicine and industry but requires numerous projections, leading to high radiation doses and long scan times.
• Conventional filtered-backprojection algorithms necessitate many views for accurate volumetric imaging.
• Sparse-view CT aims to reduce data acquisition, but reconstructing quality images from limited projections remains challenging.

Purpose of the Study:

• To investigate a super-sparse CT scanning method that significantly reduces the number of projections required for image reconstruction.
• To leverage prior CT scan data to improve image quality from extremely limited projection data.
• To demonstrate the feasibility of reconstructing high-quality CT images using a significantly reduced projection set.

Main Methods:

• Exploited image total-variation minimization, inspired by compressive sensing, to reconstruct images from sparse data.
• Utilized prior CT image information from successive scans, assuming minimal difference between scans.
• Developed and applied a super-sparse scanning approach incorporating prior data for reconstruction.

Main Results:

• Successfully reconstructed high-quality CT images from a significantly reduced number of projections (super-sparse scans).
• Demonstrated that incorporating prior scan data allows for reconstruction comparable to conventional methods, despite extreme data reduction.
• Validated the approach through both numerical simulations and experimental results.

Conclusions:

• Prior CT data significantly aids in reconstructing high-quality images from super-sparse projection sets.
• The developed method offers a viable solution for reducing radiation dose in medical CT and scan time/cost in industrial CT.
• This approach advances sparse-view CT by enabling reconstruction with unprecedentedly few projections.