Total variation superiorization schemes in proton computed tomography image reconstruction.

Area of Science:

• Medical Imaging
• Computational Physics
• Image Reconstruction

Background:

• Iterative projection algorithms are standard for proton computed tomography (pCT) reconstruction.
• Image noise increases with iterations due to proton straggling and scattering, degrading image quality.

Purpose of the Study:

• Investigate total variation superiorization (TVS) as an add-on to iterative algorithms for pCT.
• Enhance image quality and reduce noise in pCT reconstructions.

Main Methods:

• Applied two TVS schemes to the block-iterative diagonally relaxed orthogonal projections (DROP) algorithm using simulated pCT data.
• Evaluated simplified TVS schemes, omitting the proximity check, to reduce computational cost.
• Assessed spatial and density resolution using modulation transfer and contrast discrimination functions.

Main Results:

• Both TVS schemes significantly improved spatial and density resolution over standard DROP.
• Eliminating the feasibility proximity check in the once-per-block TVS scheme halved reconstruction time and reduced image noise.
• The optimized TVS approach (once per block, no proximity check) yielded the highest image quality.

Conclusions:

• Total variation superiorization (TVS) shows promise for improving low-contrast imaging in pCT.
• TVS can be effectively integrated into future pCT reconstruction workflows.