Learning a Filtered Backprojection Reconstruction Method for Photoacoustic Computed Tomography with Hemispherical...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Imaging

Background:

• Three-dimensional photoacoustic computed tomography (PACT) often uses hemispherical measurement apertures for data acquisition, resulting in half-scan data.
• Previous research confirmed the unique and stable reconstruction capabilities of half-scan data, but a closed-form reconstruction formula was lacking.

Purpose of the Study:

• To develop a semi-analytic reconstruction method for half-scan PACT data.
• To address the absence of a closed-form filtered backprojection (FBP) formula for half-scan PACT.

Main Methods:

• A novel half-scan FBP method was developed, employing a learning-based approach to approximate the unknown filtering operation.
• The method was validated using virtual imaging studies with numerical breast phantoms and a physics-based data acquisition model.
• The approach was further tested on experimental data from an in vivo breast PACT study.

Main Results:

• The half-scan FBP method successfully reconstructed accurate 3D images from half-scan PACT data.
• The learning-based approximation of the filter proved effective for accurate image reconstruction.
• The reconstruction method demonstrated robustness, maintaining accuracy with data variations.

Conclusions:

• The developed half-scan FBP method provides an effective solution for 3D PACT imaging using limited hemispherical data.
• This approach enhances the feasibility of in vivo breast imaging with PACT.
• The well-posed nature of the inverse mapping ensures reliable reconstructions even with diverse datasets.