FREQUENCY RESPONSE AND DISTORTION PROPERTIES OF RECONSTRUCTION ALGORITHMS IN COMPUTED TOMOGRAPHY

Area of Science:

• Medical Physics
• Image Reconstruction
• Computed Tomography (CT)

Background:

• Denoising reconstruction techniques in computed tomography (CT) can introduce nonlinear properties.
• These nonlinearities cause image distortion, impacting the accurate assessment of system resolution.
• Understanding and quantifying these distortions is crucial for reliable CT imaging.

Purpose of the Study:

• To decouple and investigate amplitude modulation and waveform distortion in CT reconstruction algorithms.
• To adapt existing methodologies for analyzing nonlinear image processing to CT reconstruction.
• To assess the impact of nonlinear mechanisms on CT image quality metrics.

Main Methods:

• Adapted methodology from Wells and Dobbins (2013) for CT reconstruction algorithms.
• Utilized the ASTRA Toolbox in MATLAB for reconstructing sinusoidal wave functions.
• Investigated Filtered Back Projection (FBP) with a median filter and Simultaneous Iterative Reconstruction Technique (SIRT) with a non-negative constraint.

Main Results:

• Native reconstruction algorithms exhibited nonlinear waveform distortion.
• No dependence on contrast-to-noise ratio (CNR) was observed for native algorithms.
• CT algorithms incorporating nonlinear mechanisms demonstrated a clear and specific CNR dependence.

Conclusions:

• Nonlinear waveform distortion is inherent in CT reconstruction algorithms, even without explicit nonlinear mechanisms.
• The addition of nonlinear mechanisms significantly introduces a CNR-dependent distortion.
• Analysis of distortion is essential for understanding and mitigating nonlinear effects in CT reconstruction.