SNR enhancement in brillouin microspectroscopy using spectrum reconstruction.

Area of Science:

• Spectroscopy
• Data Analysis
• Materials Science

Background:

• Brillouin spectroscopy is a powerful technique for material characterization.
• Low signal-to-noise ratios (SNRs) often compromise the reliability of Brillouin spectral analysis.
• Existing data analysis protocols struggle with SNRs below approximately 10.

Purpose of the Study:

• To investigate the effectiveness of denoising algorithms in improving Brillouin spectroscopy data.
• To enhance the accuracy and precision of determining Brillouin shifts and linewidths.
• To validate denoising algorithm performance using simulations and experimental data.

Main Methods:

• Exploitation of maximum entropy reconstruction (MER) and wavelet analysis (WA) for signal denoising.
• Quantification of algorithm performance via Monte-Carlo simulations.
• Benchmarking against the Cramér-Rao lower bound for estimation accuracy.
• Application of denoising to experimental Brillouin spectra of distilled water.

Main Results:

• MER and WA significantly improve the accuracy and precision of Brillouin shift and linewidth determination.
• Superior estimation results were achieved even at low SNRs (≥ 1).
• Denoising enabled accurate extraction of the speed of sound in water from experimental data.
• Experimental and theoretical speed of sound values agreed within ±1% at unity SNR.

Conclusions:

• Maximum entropy reconstruction and wavelet analysis are effective denoising strategies for low-SNR Brillouin spectroscopy.
• These algorithms enhance the reliability and precision of spectral parameter extraction.
• Denoising facilitates accurate material property measurements, such as the speed of sound, even under challenging experimental conditions.