Artifact reduction in photoacoustic images by generating virtual dense array sensor from hemispheric sparse array...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Artificial Intelligence

Background:

• Vascular imaging is crucial for disease diagnosis and surgical planning.
• Photoacoustic imaging offers noninvasive, high-resolution visualization of blood vessels.
• Hemispherical array sensors are ideal for diverse vascular imaging, but sparse arrays cause artifacts.

Purpose of the Study:

• To reduce artifacts in photoacoustic images caused by sparse hemispherical array sensors.
• To enhance the visualization of deep blood vessels using deep learning.
• To develop a computationally efficient method for artifact reduction.

Main Methods:

• A 2D convolutional neural network (CNN) was employed to generate virtual sensor signals.
• Virtual sensors were strategically placed between real sensors in a spiral pattern.
• Photoacoustic images were reconstructed using both real and generated virtual sensor signals.

Main Results:

• Significant reduction in image artifacts was observed compared to using only real sensor data.
• The proposed method demonstrated effectiveness on both simulation and human palm measurement data.
• Reconstructed images showed improved clarity and reduced artifact presence.

Conclusions:

• The deep learning approach effectively minimizes artifacts in photoacoustic imaging.
• The method enables clearer visualization of deep blood vessels.
• Processing times are suitable for clinical measurements, indicating practical applicability.