Photoacoustic image improvement based on a combination of sparse coding and filtering

Area of Science:

• Biomedical Imaging
• Optical Physics
• Signal Processing

Background:

• Photoacoustic imaging (PAI) is valuable for diagnostics but suffers from low signal-to-noise ratio (SNR) due to tissue noise.
• Traditional noise reduction techniques like frame averaging degrade temporal resolution, limiting PAI applications.
• Improving SNR is crucial for accurate interpretation and diagnostic efficacy in PAI.

Purpose of the Study:

• To develop and evaluate a novel denoising method for photoacoustic signals.
• To enhance image quality in PAI by improving SNR without compromising temporal resolution.
• To provide an effective tool for clearer PAI, particularly for imaging deeper structures.

Main Methods:

• A novel denoising approach combining low-pass filtering and sparse coding (LPFSC) was proposed.
• The method models photoacoustic signals as a sum of low-frequency and sparse components.
• An optimization process using a hybrid alternating direction method of multipliers was employed for noise reduction.

Main Results:

• The LPFSC method demonstrated a 26% improvement in peak SNR for simulated data compared to frame averaging.
• Experimental results showed a 63% increase in contrast-to-noise ratio and a 33% improvement in structural similarity index.
• These improvements were observed for objects at depths up to 20 mm in a porcine tissue phantom.

Conclusions:

• The LPFSC method effectively denoises photoacoustic signals, significantly enhancing reconstructed image quality.
• This technique is particularly beneficial for imaging at greater depths.
• The LPFSC approach improves PAI without limiting image acquisition speed, offering a valuable advancement for diagnostic applications.