Denoising Raman spectra by Wiener estimation with a numerical calibration dataset

Area of Science:

• Spectroscopy
• Biomedical Optics
• Data Processing

Background:

• Raman spectral data processing often requires manual parameter tuning for effective denoising.
• Existing denoising techniques can be sensitive to signal-to-noise ratios (SNR), impacting reliability.
• Current Wiener estimation methods necessitate experimental calibration data.

Purpose of the Study:

• To develop an automated Raman spectra denoising method.
• To reduce user dependency in spectral data processing.
• To improve denoising performance, especially in low SNR conditions.

Main Methods:

• A novel denoising approach combining spectral integration and Wiener estimation.
• Utilized a numerical calibration dataset, eliminating the need for experimental measurements.
• Tested on diverse samples: phantom, human fingernail, and leukemia cells.

Main Results:

• The proposed method demonstrates significantly reduced sensitivity to parameter selection compared to moving-average and Savitzky-Golay filters.
• Achieved comparable or superior denoising performance, particularly in low SNR scenarios.
• Successfully applied to various sample types, showcasing versatility.

Conclusions:

• The new spectral integration and Wiener estimation method offers an efficient and robust solution for Raman spectra denoising.
• This approach minimizes user interaction and enhances reliability across different SNR levels.
• Provides a valuable tool for accurate analysis of complex spectral data.