Comprehensive chemometric and machine learning methods for determination of synthetic dyes adulteration in ceylon tea via NIR spectroscopy and hyperspectral imaging

Menevseoglu A., Gunes N., AĞÇAM E., TURGUT S. S., Pérez-Marín D.

Microchemical Journal, cilt.218, 2025 (SCI-Expanded) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 218
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1016/j.microc.2025.115697
  • Dergi Adı: Microchemical Journal
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, BIOSIS, CAB Abstracts, Chemical Abstracts Core, Chimica, Food Science & Technology Abstracts, Index Islamicus, Veterinary Science Database
  • Anahtar Kelimeler: adulteration, chemometrics, HSI, NIR, synthetic dye, tea
  • Çukurova Üniversitesi Adresli: Evet

Özet

The increasing prevalence of synthetic dye adulteration in Ceylon black tea presents a significant challenge to food safety and consumer trust. This study aimed to develop a rapid, non-destructive detection method for synthetic dyes in tea using portable Near-Infrared (NIR) spectroscopy and Hyperspectral Imaging (HSI) combined with advanced chemometric and machine learning techniques. A total of 100 tea samples were analyzed, and synthetic dyes, including Allura Red and Carmoisine, were detected in a subset of samples using HPLC as a reference method. Chemometric techniques such as SIMCA, PLS-DA, kNN, MCR, and ALS demonstrated high accuracy in distinguishing authentic and adulterated samples. Conditional Entropy (CE)-based feature extraction identified key wavelengths critical for classification, significantly enhancing model performance. Machine learning models, particularly Support Vector Machines (SVM) and Ensemble methods, achieved classification accuracies of up to 99.7 % for NIR and 100 % for HSI data. HSI consistently outperformed NIR in classification due to its ability to capture both spatial and spectral information, with superior interclass distance and variance explained. The broader scanning area of HSI (∼25 cm2) provided richer datasets compared to NIR (∼2.27 cm2). These findings underscore the potential of NIR and HSI as a robust, field-deployable tool for rapid screening of synthetic dye adulteration in tea. This study demonstrated the effectiveness of integrating spectroscopy with chemometrics and machine learning for food authentication and safety, offering an efficient alternative to traditional, labor-intensive methods. The proposed approaches provided critical insights into developing portable, real-time detection systems for ensuring the integrity of tea products in the global market.