Determination of the lactose content in low-lactose milk using Fourier-transform infrared spectroscopy (FTIR) and convolutional neural network

被引:11
作者
Ribeiro, Daniela C. S. Z. [1 ]
Neto, Habib Asseiss [2 ]
Lima, Juliana S. [1 ]
de Assis, Debora C. S. [1 ]
Keller, Kelly M. [1 ]
Campos, Sergio V. A. [3 ]
Oliveira, Daniel A. [4 ]
Fonseca, Leorges M. [1 ]
机构
[1] Univ Fed Minas Gerais UFMG, Sch Vet Med, Belo Horizonte, MG, Brazil
[2] Fed Inst Mato Grosso Sul, Tres Lagoas, MS, Brazil
[3] Univ Fed Minas Gerais UFMG, Dept Comp Sci, Belo Horizonte, MG, Brazil
[4] Ezequiel Dias Fdn FUNED MG, Belo Horizonte, MG, Brazil
关键词
Low; -lactose; Milk; Artificial neural network; Convolutional neural network; Deep learning; QUANTIFICATION;
D O I
10.1016/j.heliyon.2023.e12898
中图分类号
O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];
学科分类号
07 ; 0710 ; 09 ;
摘要
Demand for low lactose milk and milk products has been increasing worldwide due to the high number of people with lactose intolerance. These low lactose dairy foods require fast, low-cost and efficient methods for sugar quantification. However, available methods do not meet all these requirements. In this work, we propose the association of FTIR (Fourier Transform Infrared) spectroscopy with artificial intelligence to identify and quantify residual lactose and other sugars in milk. Convolutional neural networks (CNN) were built from the infrared spectra without preprocessing the data using hyperparameter adjustment and saliency map. For the quantitative prediction of the sugars in milk, a regression model was proposed, while for the qualitative assessment, a classification model was used. Raw, pasteurized and ultra-high temperature (UHT) milk was added with lactose, glucose, and galactose in six concentrations (0.1-7.0 mg mL-1) and, in total, 432 samples were submitted to convolutional neural network. Accuracy, precision, sensitivity, specificity, root mean square error, mean square error, mean absolute error, and coefficient of determination (R2) were used as evaluation parameters. The algorithms indicated a predictive capacity (accuracy) above 95% for classification, and R2 of 81%, 86%, and 92% for respectively, lactose, glucose, and galactose quantification. Our results showed that the associa-tion of FTIR spectra with artificial intelligence tools, such as CNN, is an efficient, quick, and low-cost methodology for quantifying lactose and other sugars in milk.
引用
收藏
页数:12
相关论文
共 38 条
  • [1] Abadi M, 2016, PROCEEDINGS OF OSDI'16: 12TH USENIX SYMPOSIUM ON OPERATING SYSTEMS DESIGN AND IMPLEMENTATION, P265
  • [2] Acquaro Jr V.R., 2013, SCI CHROMATOGR, V5, P137, DOI 10.4322/
  • [3] Alpaydin E., 2020, Introduction to Machine Learning, V4th
  • [4] On the utilization of deep and ensemble learning to detect milk adulteration
    Asseiss Neto, Habib
    Tavares, Wanessa L. F.
    Ribeiro, Daniela C. S. Z.
    Alves, Ronnie C. O.
    Fonseca, Leorges M.
    Campos, Sergio V. A.
    [J]. BIODATA MINING, 2019, 12 (1)
  • [5] Brasil: Ministerio da Saude/Age^ncia Nacional de Vigila^ncia Sanitaria RDC N◦136, 2017, 136 RDC MIN SAUD AG
  • [6] Brunt Kommer, 2020, J AOAC Int, DOI 10.1093/jaocint/qsaa092
  • [7] Chollet F, 2015, KERAS
  • [8] Chollet F., 2021, DEEP LEARNING PYTHON
  • [9] Use of FTIR-ATR Spectroscopy Combined with Multivariate Analysis as a Screening Tool to Identify Adulterants in Raw Milk
    Conceicao, Daniele G.
    Goncalves, Ben-Hur R. F.
    da Hora, Fernanda F.
    Faleiro, Amanda S.
    Santos, Leandro S.
    Ferrao, Sibelli P. B.
    [J]. JOURNAL OF THE BRAZILIAN CHEMICAL SOCIETY, 2019, 30 (04) : 780 - 785
  • [10] Quantification of lactose using ion-pair RP-HPLC during enzymatic lactose hydrolysis of skim milk
    Erich, Sarah
    Anzmann, Theresa
    Fischer, Lutz
    [J]. FOOD CHEMISTRY, 2012, 135 (04) : 2393 - 2396