Spiral scanning and self-supervised image reconstruction enable ultra-sparse sampling multispectral photoacoustic...

Area of Science:

• Biomedical Imaging
• Optical Imaging
• Photoacoustic Imaging

Background:

• Multispectral photoacoustic tomography (PAT) offers high-contrast, non-invasive imaging and molecular functional information.
• Existing multispectral PAT systems face high hardware costs and computational demands due to numerous detectors.
• Addressing these limitations is crucial for advancing molecular imaging applications.

Purpose of the Study:

• To introduce an ultra-sparse spiral sampling strategy for multispectral PAT (U3S-PAT).
• To reduce hardware complexity and computational load in multispectral PAT systems.
• To enable efficient volumetric molecular imaging of dynamic biological processes.

Main Methods:

• Developed U3S-PAT employing a sparse, ring-shaped transducer with simultaneous rotation and translation for spiral scanning.
• Implemented multispectral angle-interlaced sampling within the spiral pattern.
• Utilized a self-supervised learning method for image reconstruction to address ill-conditioning and incorporate shared structural information.

Main Results:

• U3S-PAT achieved comparable reconstruction and spectral unmixing accuracy to dense sampling methods, even at a 1/30 sparse sampling rate.
• Simulations on a commercial PAT system and in vivo animal experiments validated the U3S-PAT strategy.
• The proposed method demonstrated significant reduction in 3D multispectral scanning time.

Conclusions:

• U3S-PAT offers a cost-effective and computationally efficient solution for multispectral PAT.
• The strategy enables high-accuracy molecular imaging with reduced data acquisition time.
• U3S-PAT holds potential for real-time, volumetric molecular imaging of biological activities.