A rapid, non-destructive, and accurate method for identifying citrus granulation using Raman spectroscopy and machine...

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Spectroscopy

Background:

• Citrus fruit quality is impacted by juice sac granulation during storage.
• This condition presents a significant challenge to the citrus industry, affecting marketability and leading to economic losses.

Purpose of the Study:

• To develop a rapid, non-destructive, and precise method for detecting citrus granulation.
• To utilize Raman spectroscopy and machine learning for accurate granulation assessment.

Main Methods:

• Analysis of 969 Raman spectral data points from granulated and non-granulated citrus samples.
• Application of logistic regression, decision tree, and partial least squares discriminant analysis.
• Refinement of models using principal component analysis, successive projection algorithm, and competitive adaptive reweighted sampling (CARS).

Main Results:

• Identification of characteristic Raman peaks (1580 and 1661 cm⁻¹) indicative of granulation, linked to water, ferulic acid, and sugar content differences.
• Partial least squares discriminant analysis achieved high accuracy (0.997), recall (0.994), and F-fraction (0.996).
• A combined second derivative-CARS-partial least squares discriminant analysis model demonstrated 100% accuracy in test sets.

Conclusions:

• The proposed Raman spectroscopy and machine learning approach offers a robust and reliable method for assessing citrus fruit quality and detecting granulation.
• This technology can be applied for in-line screening of citrus during processing, minimizing waste and economic losses.
• Provides technical support for classifying citrus crop quality, enhancing industry standards.