Discriminating forchlorfenuron-treated kiwifruits using a portable spectrometer and Vis/NIR diffuse transmittance spectroscopy technology. Issue 28 (11th July 2017)

• Record Type: Journal Article
• Authors: Liu, Dayang; Li, Qianqian; Li, Weiqiang; Yang, Biao; Guo, Wenchuan
• Is Part Of: Analytical methods. Volume 9:Issue 28(2017)
• Page Start: 4207
• Page End: 4214
• Publication Date: 2017-07-11
• Subjects: Chemistry, Analytic -- Periodicals; Analytical biochemistry -- Periodicals; Chemical laboratories -- Standards -- Periodicals; 543.1905
• DOI: 10.1039/c7ay00832e
• Languages: English
• ISSNs: 1759-9660

Abstract

Abstract : To explore the feasibility of low-cost miniaturized spectrometers in discriminating forchlorfenuron treated kiwifruits, a visible and near infrared (Vis/NIR) diffuse transmittance spectroscopy system was constructed by using a miniaturized charge coupled device (CCD) spectrometer in the wavelength range of 400–1100 nm and a self-developed optical fiber probe. Abstract : To explore the feasibility of low-cost miniaturized spectrometers in discriminating forchlorfenuron treated kiwifruits, a visible and near infrared (Vis/NIR) diffuse transmittance spectroscopy system was constructed by using a miniaturized charge coupled device (CCD) spectrometer in the wavelength range of 400–1100 nm and a self-developed optical fiber probe. Two hundred and forty spectra were acquired from two positions about 180° apart around the equatorial position of 120 'Xuxiang' kiwifruits (including 60 forchlorfenuron-treated kiwifruits and 60 untreated kiwifruits). The spectra in the range of 500–950 nm including 2386 wavelengths were used in this study. The successive projections algorithm (SPA) was used to select effective wavelengths (EWs). Support vector machine (SVM), extreme learning machine (ELM), and partial least squares discriminant analysis (PLS-DA) modeling methods were used to build discrimination models. 40 EWs were selected from 2386 full wavelengths by using the SPA. The average accuracy rates of each developed model were higher than 98.1% and 88.8% for the calibration set and prediction set, respectively. ELM models achieved better calibration and prediction performance than SVM and PLS-DA models. The SPA-ELM model, whose discrimination accuracy rate was 100% for the calibration set and 95.0% for the prediction set, was the best model for the discrimination of forchlorfenuron-treated kiwifruits from untreated ones. This research indicates that portable detectors with high accuracy in discriminating forchlorfenuron-treated kiwifruits could be developed by using cheap and miniaturized CCD spectrometers.