Discrimination of Grape Seeds Using Laser-Induced Breakdown Spectroscopy in Combination with Region Selection and...

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Biotechnology

Background:

• Grape pomace, a wine industry byproduct, contains valuable grape seeds.
• Grape seeds are rich in bioactive compounds, but their nutritional properties vary by cultivar.
• Effective identification of grape seed cultivars is crucial for their valorization.

Purpose of the Study:

• To develop and evaluate a rapid and accurate method for identifying grape seed cultivars.
• To explore the relationship between spectral responses and cultivar information using chemometrics and deep learning.
• To demonstrate the potential of laser-induced breakdown spectroscopy (LIBS) for grape seed analysis.

Main Methods:

• Spectral profiles of grape seeds from three cultivars were acquired using laser-induced breakdown spectroscopy (LIBS).
• Conventional supervised classification methods and a one-dimensional convolutional neural network (CNN) were employed.
• Interval partial least squares (iPLS) was used to select relevant spectral regions.

Main Results:

• The interval partial least squares (iPLS) algorithm identified a spectral region (402.74-426.87 nm) relevant for elemental composition.
• The deep learning CNN model achieved optimal performance, with 100% accuracy on the calibration set and 96.7% on the prediction set.
• The CNN model utilizing full spectra outperformed models based on selected spectral regions.

Conclusions:

• LIBS is a reliable, rapid, and accurate technique for identifying grape seed cultivars.
• This approach facilitates the utilization of grape seeds with desirable nutritional properties, minimizing waste.
• The study supports the valorization of grape seeds in the wine industry.