Improving spatial resolution with an edge-enhancement model for low-dose propagation-based X-ray phase-contrast...

Area of Science:

• Medical Imaging
• Physics
• Computational Science

Background:

• Propagation-based X-ray phase-contrast computed tomography (PB-PCCT) is valuable for imaging low-contrast tissues.
• Traditional phase retrieval methods, like homogenous transport of intensity equation (TIE-Hom), often reduce spatial resolution while enhancing signal-to-noise ratio.
• This trade-off limits the diagnostic potential of PB-PCCT in scenarios requiring high detail.

Purpose of the Study:

• To develop an enhanced iterative algorithm for PB-PCCT that improves reconstructed spatial resolution and suppresses noise.
• To address the limitations of conventional phase retrieval methods in achieving both high resolution and low noise.
• To enable better visualization of fine details in low-dose or sparse-view PB-PCCT applications.

Main Methods:

• A novel approach combining weighted summed measured phase contrast projections (with edge enhancement) and TIE-Hom retrieved phase projections.
• Integration of this merged data into an adaptive steepest descent projections onto convex sets (ASD-POCS) iterative algorithm.
• Introduction of an additive median root prior to enhance model accuracy and an additive median root prior to improve model accuracy.

Main Results:

• The proposed enhanced ASD-POCS algorithm demonstrated superior performance over conventional ASD-POCS in numerical and experimental datasets.
• Significant improvements were observed in spatial resolution metrics (e.g., 66.48% increase in MTF50%) and noise reduction (e.g., 35.25% decrease in f50%).
• Experimental validation showed an average increase of 67.35% in contrast-to-noise ratio (CNR) and enhanced edge sharpness.

Conclusions:

• The developed algorithm effectively balances spatial resolution and noise performance in PB-PCCT.
• This method offers a significant advancement for low-dose and sparse-view PB-PCCT, improving image quality for distinguishing subtle tissue contrasts.
• The flexible control over antagonistic hyper-parameters allows for tailored optimization based on specific imaging requirements.