Mitigating the Limited View Problem in Photoacoustic Tomography for a Planar Detection Geometry by Regularized...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Imaging

Background:

• Photoacoustic tomography (PAT) with planar detection suffers from the limited-view problem due to finite aperture.
• One-step reconstruction algorithms yield compromised image quality, including streaking artifacts, reduced contrast, and poor signal-to-noise ratio.
• Addressing these limitations is crucial for accurate PAT imaging.

Purpose of the Study:

• To evaluate model-based iterative reconstruction (MBIR) approaches for improving PAT image quality.
• To compare MBIR methods with total variation (TV) regularization against one-step methods.
• To investigate the impact of regularization parameters and Bregman iterations on image reconstruction.

Main Methods:

• In-silico, phantom, ex vivo, and in vivo data were used to assess reconstruction algorithms.
• Least squares minimization with and without TV regularization formed the basis of MBIR.
• Bregman iterations were explored to correct amplitude bias introduced by TV regularization.

Main Results:

• MBIR methods demonstrated superior image quality compared to one-step methods.
• Improvements were observed in signal-to-artifact ratio, signal-to-noise ratio, amplitude accuracy, and spatial fidelity.
• The study analyzed the influence of the regularization parameter on feature scale and amplitude distribution.

Conclusions:

• Model-based iterative reconstruction significantly enhances photoacoustic image quality in limited-view scenarios.
• MBIR offers a robust solution for mitigating artifacts and improving diagnostic accuracy in PAT.
• This work provides insights for applying MBIR in practical photoacoustic imaging systems with planar detectors.