A Photoacoustic Imaging Algorithm Based on Regularized Smoothed L₀ Norm Minimization

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Sparse reconstruction is crucial for efficient photoacoustic imaging (PAI).
• Compressed sensing (CS) offers potential for high-quality PAI image reconstruction from sparse signals.
• Existing CS algorithms face challenges with noise and accuracy.

Purpose of the Study:

• To introduce a novel CS-based algorithm, ReSL0, for PAI image reconstruction.
• To enhance error tolerance and noise immunity in PAI reconstruction.
• To evaluate the performance of ReSL0 against existing methods.

Main Methods:

• Developed a CS-based error-tolerant regularized smooth L0 (ReSL0) algorithm.
• Reconstructed simulated datasets from three phantoms.
• Validated the algorithm using a real experimental dataset from an agar phantom.
• Compared ReSL0 with L0, L1, and TV norm-based CS algorithms.

Main Results:

• ReSL0 demonstrated a good balance between reconstruction quality and efficiency.
• The algorithm showed improved immunity to noise compared to other methods.
• Peak Signal-to-Noise Ratio (PSNR) of ReSL0 reconstructions was superior to comparative algorithms.
• ReSL0 notably enhanced reconstruction quality in the presence of noisy measurements.

Conclusions:

• The ReSL0 algorithm is effective for PAI image reconstruction.
• It offers significant advantages in handling noisy data.
• ReSL0 provides a robust and efficient solution for PAI, balancing quality and computational performance.