Spatiospectral denoising framework for multispectral optoacoustic imaging based on sparse signal representation

Area of Science:

• Optoacoustic imaging
• Biomedical optics
• Signal processing

Background:

• Dynamic multispectral optoacoustic imaging faces challenges with low signal-to-noise ratio, necessitating signal averaging.
• This limitation hinders the imaging rate and overall dynamic capabilities of the technology.

Purpose of the Study:

• To develop an advanced denoising method for multispectral optoacoustic imaging.
• To overcome the limitations imposed by low signal-to-noise ratio and enable faster imaging.

Main Methods:

• A novel denoising approach exploiting signal sparsity in both spatial and spectral domains.
• Utilizes a combined wavelet-Karhunen-Loève representation for effective noise suppression.
• Thresholding is applied to the sparse representation to remove noise while preserving signal integrity.

Main Results:

• Demonstrated significant improvements in image quality for in vivo mouse data.
• Successfully reduced common noise sources, including time-varying parasitic signals and white noise.
• Outperformed existing denoising strategies that do not leverage multispectral information.

Conclusions:

• The proposed denoising methodology effectively suppresses noise with minimal signal loss.
• Offers a promising solution for enhancing the dynamic capabilities of multispectral optoacoustic imaging.
• Maintains high image quality, paving the way for advanced applications.