Deep-E Enhanced Photoacoustic Tomography Using Three-Dimensional Reconstruction for High-Quality Vascular Imaging

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Artificial Intelligence

Background:

• Linear-array photoacoustic computed tomography (PACT) is valuable for vascular imaging due to cost-effectiveness and ultrasound compatibility.
• Linear-array transducers exhibit limitations in three-dimensional (3D) imaging, primarily due to poor elevation resolution.

Purpose of the Study:

• To develop and validate a deep learning-assisted data processing algorithm for enhancing image quality in linear-array PACT.
• To improve elevation resolution and the visualization of deep vascular structures in PACT imaging.

Main Methods:

• A novel deep learning algorithm was developed, training two separate networks on 2D and 3D reconstructed PACT data.
• Image data from both 2D and 3D training were fused to leverage features from both algorithms.
• The algorithm's efficacy was validated using numerical simulations and in vivo experiments.

Main Results:

• The deep learning approach significantly improved elevation resolution compared to conventional methods.
• The algorithm accurately recovered the true size of imaged objects.
• Enhanced visualization of deep vessels was achieved, demonstrating improved PACT performance.

Conclusions:

• The proposed deep learning-assisted algorithm effectively addresses the elevation resolution limitations of linear-array PACT.
• This technique offers improved accuracy in object size recovery and enhanced visualization of deep vascular networks.
• The approach holds significant potential for translational imaging applications requiring detailed vascular feature visualization.