Few-view image reconstruction with fractional-order total variation

Area of Science:

• Medical Imaging
• Computational Imaging
• Applied Mathematics

Background:

• Few-view computed tomography (CT) reconstruction is challenging due to limited data.
• Traditional total variation minimization methods can cause image over-smoothing.
• Existing methods like FBP, TV-POCS, and STH have limitations in few-view scenarios.

Purpose of the Study:

• To develop a novel CT reconstruction method for few-view problems.
• To address the over-smoothing issue inherent in traditional total variation methods.
• To improve image reconstruction quality using fractional calculus.

Main Methods:

• Proposed a fractional-order total variation-based image reconstruction model.
• Utilized fractional-order total variation, considering more neighboring voxels for adaptive weighting.
• Developed a discretization scheme for the fractional-order model.

Main Results:

• The fractional-order total variation method effectively suppresses the over-smoothing effect.
• Numerical and clinical experiments validated the method's performance.
• Achieved superior results compared to filtered back projection (FBP), TV-POCS, and STH.

Conclusions:

• The proposed fractional-order total variation method offers improved CT image reconstruction for few-view problems.
• This novel approach enhances image detail and quality.
• Demonstrated superior performance against conventional and existing advanced reconstruction techniques.