Spectroscopic photoacoustic denoising framework using hybrid analytical and data-free learning method.

Area of Science:

• Medical imaging
• Biomedical optics
• Photoacoustic imaging

Background:

• Spectroscopic photoacoustic (sPA) imaging quantifies chromophores using spectral absorption.
• PA imaging is prone to noise, reducing signal-to-noise ratio (SNR) and spectral unmixing accuracy.
• Existing denoising methods have limitations, including impracticality for dynamic imaging and dependence on training data.

Purpose of the Study:

• To develop a novel denoising framework for spectroscopic photoacoustic imaging.
• To address challenges of noise, low SNR, and spectral information preservation in PA imaging.
• To create a clinically adaptable method that does not require extensive training data.

Main Methods:

• Proposed the Spectroscopic Photoacoustic Denoising (SPADE) framework.
• Integrated a data-free learning-based method with an efficient BM3D-based analytical approach.
• Validated SPADE through simulation, phantom, in vivo, and ex vivo studies.

Main Results:

• SPADE improved image SNR by over 15 dB in high noise scenarios.
• Preserved spectral information with a correlation coefficient (R) greater than 0.8.
• Outperformed conventional denoising methods, particularly in low SNR conditions.

Conclusions:

• SPADE offers a promising solution for noise reduction in quantitative PA imaging.
• The framework effectively preserves spectral integrity and functional information.
• SPADE enhances the accuracy and clinical applicability of PA imaging where noise is a concern.