ARCHER. A new algorithm for automatic removal of cosmic spikes and saturated pixels in hyperspectral Raman...

Area of Science:

• Spectroscopy
• Data Science
• Imaging

Background:

• Cosmic spikes are a significant challenge in Raman spectroscopy preprocessing.
• Existing software methods for cosmic spike removal are often ineffective or poorly documented.
• Accurate preprocessing is crucial for reliable multivariate analysis of hyperspectral Raman data.

Purpose of the Study:

• To introduce ARCHER, a novel algorithm for automatic cosmic spike correction in Hyperspectral Raman Imaging.
• To enable effective detection and removal of both cosmic spikes and saturated pixels.
• To optimize algorithm parameters for enhanced performance.

Main Methods:

• ARCHER combines multiple steps and factorizations for robust cosmic spike detection.
• The algorithm incorporates saturated pixel detection.
• Parameter optimization was performed using D-optimal design, validated on artificial and real hyperspectral images.

Main Results:

• ARCHER demonstrates effective and automatic detection and removal of cosmic spikes.
• The algorithm successfully identifies saturated pixels, essential for subsequent analysis.
• Validation across various complexity levels confirmed algorithm robustness.

Conclusions:

• ARCHER provides an effective solution for cosmic spike and saturated pixel removal in Hyperspectral Raman Imaging.
• The algorithm enhances the preprocessing pipeline for Raman spectroscopy data.
• This tool improves the reliability of downstream multivariate analyses.