Noise texture and signal detectability in propagation-based x-ray phase-contrast tomography

Area of Science:

• Medical Imaging
• Image Reconstruction
• Quantitative Analysis

Background:

• X-ray phase-contrast tomography (PCT) is an emerging 3D imaging technique.
• It reconstructs an object's x-ray refractive index distribution.
• Statistical properties of PCT images are not well understood.

Purpose of the Study:

• To quantitatively investigate noise propagation in PCT image reconstruction.
• To characterize noise texture and its dependence on imaging geometry.
• To understand the impact of noise on signal detection performance.

Main Methods:

• Derived explicit expressions for autocovariance of reconstructed absorption and refractive index images.
• Employed statistical detection theory to analyze imaging geometry effects.
• Compared analytical formulas with numerical and empirical values.

Main Results:

• Excellent agreement between analytical and empirical autocovariance values.
• Reconstructed refractive images exhibit high spatial correlation; absorption images do not.
• Detection performance increases with detector-plane spacing.

Conclusions:

• First quantitative investigation of noise propagation in PCT.
• Spatial correlation in refractive images stems from a Fourier space singularity.
• Statistical analysis aids in optimizing PCT for specific applications.