Progressive dual-branch adversarial diffusion for sparse-view photoacoustic tomography reconstruction.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Imaging

Background:

• Photoacoustic computed tomography (PACT) offers high optical contrast and deep tissue penetration.
• Sparse-view PACT acquisitions, due to hardware constraints, lead to significant image artifacts.
• Existing reconstruction methods struggle with quality and efficiency under sparse data.

Purpose of the Study:

• To develop an efficient and high-quality reconstruction framework for sparse-view PACT.
• To address the challenge of severe artifacts in PACT imaging with limited angular sampling.
• To improve the inference speed of PACT reconstruction algorithms.

Main Methods:

• Proposed ND-net, a progressive dual-branch adversarial diffusion framework.
• Implemented a two-stage approach: artifact reconstruction followed by full-view refinement.
• Utilized flexible reverse transitions for large-step diffusion sampling with few iterations.

Main Results:

• ND-net demonstrated superior reconstruction quality compared to existing methods on simulated and experimental data.
• Achieved significant artifact reduction in sparse-view PACT.
• Showcased efficient inference capabilities with only four reverse iterations.

Conclusions:

• ND-net effectively improves sparse-view PACT reconstruction quality.
• The framework offers a significant advancement in efficient and high-fidelity PACT imaging.
• This method holds promise for clinical translation of PACT technology.