A Study on the Screening of Anti-Inflammatory Drug Diclofenac Sodium in Dietary Supplements by Near Infrared Hyperspectral Imaging

Cheng WANG; Hang YU; Wei-rong YAO; Yu-liang CHENG; Ya-hui GUO; He QIAN; Zhi-qiang TAN; Yun-fei XIE

doi:10.3964/j.issn.1000-0593(2021)02-0592-07

Journals >Spectroscopy and Spectral Analysis >Volume 41 >Issue 2 >Page 592 > Article

Spectroscopy and Spectral Analysis
Vol. 41, Issue 2, 592 (2021)

A Study on the Screening of Anti-Inflammatory Drug Diclofenac Sodium in Dietary Supplements by Near Infrared Hyperspectral Imaging

Cheng WANG^1、1, Hang YU^1、1, Wei-rong YAO^1、1, Yu-liang CHENG^1、1, Ya-hui GUO^1、1, He QIAN^1、1, Zhi-qiang TAN^1、1, and Yun-fei XIE^1、1

Author Affiliations

¹[in Chinese]

¹1. School of Food Science and Technology, Jiangnan University, Wuxi 214122, China

show less

DOI: 10.3964/j.issn.1000-0593(2021)02-0592-07 Cite this Article

Cheng WANG, Hang YU, Wei-rong YAO, Yu-liang CHENG, Ya-hui GUO, He QIAN, Zhi-qiang TAN, Yun-fei XIE. A Study on the Screening of Anti-Inflammatory Drug Diclofenac Sodium in Dietary Supplements by Near Infrared Hyperspectral Imaging[J]. Spectroscopy and Spectral Analysis, 2021, 41(2): 592 Copy Citation Text

show less

Mean original spectra of Diclofenac Sodium, Anti-inflammatory dietary supplements (KB), different concentrations of diclofenac sodium compounds (No.1—20) in inflammation-reducing dietary supplements

Fig. 1. Mean original spectra of Diclofenac Sodium, Anti-inflammatory dietary supplements (KB), different concentrations of diclofenac sodium compounds (No.1—20) in inflammation-reducing dietary supplements

Download full size | View in the Article

Hyperspectral images of 0.01%~20% Diclofenac sodium in anti-inflammatory dietary supplements (KB), anti-inflammatory dietary supplements and Diclofenac sodium (Diclofenac) at 1 675 nm

Fig. 2. Hyperspectral images of 0.01%~20% Diclofenac sodium in anti-inflammatory dietary supplements (KB), anti-inflammatory dietary supplements and Diclofenac sodium (Diclofenac) at 1 675 nm

Download full size | View in the Article

Fig. 3. Selection of the optimal spectral bands by weight ratio of β regression coefficients in the partial least squares regression (PLSR) model

Download full size | View in the Article

Model	Pretreatment	Calibration			Validation		Prediction
Model	Pretreatment	Lvs	R²	RMSEC	R²	RMSECV	R²	RMSEP	SEP	Bias
PLSR	RAW	6	0.980 4	0.009 4	0.972 9	0.011 1	0.946 8	0.012 9	0.012 6	0.003 8
	Moving Average	6	0.979 6	0.009 5	0.973 1	0.011 1	0.933 2	0.014 5	0.013 3	0.006 2
	Gaussian Filter	6	0.980 2	0.009 4	0.973 5	0.011 0	0.892 6	0.013 4	0.012 7	0.004 5
	Median Filter	6	0.980 0	0.009 5	0.974 5	0.010 9	0.947 3	0.012 9	0.012 7	0.003 0
	SG-Smoothing	6	0.979 8	0.009 5	0.975 3	0.010 8	0.943 3	0.013 4	0.013 1	0.004 0
	Normalize	4	0.988 0	0.007 3	0.983 9	0.008 4	0.971 2	0.009 5	0.008 2	0.005 0
	Baseline	6	0.981 8	0.009 0	0.976 0	0.010 7	0.949 9	0.012 6	0.012 2	0.003 8
	SNV	5	0.984 0	0.008 5	0.978 4	0.009 9	0.956 9	0.011 7	0.011 0	0.004 4
	MSC	5	0.984 2	0.008 4	0.979 6	0.009 7	0.957 3	0.011 6	0.010 9	0.004 3

Table 1. PLSR model based on full spectral wavelength

View in the Article

Model	Pretreatment	Calibration			Validation		Prediction
Model	Pretreatment	Lvs	R²	RMSEC	R²	RMSECV	R²	RMSEP	SEP	Bias
PCR	RAW	5	0.955 0	0.014 2	0.953 7	0.014 9	0.907 3	0.018 0	0.018 2	0.001 5
	Moving Average	5	0.953 0	0.014 5	0.947 8	0.015 3	0.897 0	0.019 9	0.019 3	0.006 2
	Gaussian Filter	5	0.954 5	0.014 3	0.951 8	0.015 0	0.874 2	0.018 4	0.018 5	0.002 4
	Median Filter	5	0.954 4	0.014 3	0.949 4	0.015 1	0.892 6	0.018 2	0.018 2	0.003 1
	SG-Smoothing	5	0.953 5	0.014 4	0.948 0	0.015 3	0.894 6	0.018 5	0.018 8	0.001 1
	Normalize	4	0.985 1	0.022 6	0.984 2	0.023 4	0.891 4	0.024 4	0.024 7	0.000 0
	Baseline	4	0.952 4	0.014 6	0.946 8	0.015 6	0.811 6	0.018 4	0.018 6	0.001 8
	SNV	4	0.962 5	0.013 0	0.957 5	0.013 8	0.892 0	0.018 7	0.019 0	0.000 6
	MSC	5	0.992 3	0.016 0	0.991 8	0.016 9	0.888 7	0.018 4	0.018 2	0.004 1

Table 2. PCR model based on full spectral wavelength

View in the Article

Model	pretreatment	Calibration		Validation		Prediction
Model	pretreatment	R²	RMSEC	R²	RMSECV	R²(Pearson)	RMSEP	SEP	Bias
MLR	RAW	0.998	0.004	0.991 8	0.006 1	0.987 1	0.006 4	0.006 5	0
	Moving Average	0.998 5	0.003 6	0.993 3	0.005 4	0.972 1	0.023 7	0.011 9	0.020 6
	Gaussian Filter	0.998 3	0.003 9	0.992 0	0.006 0	0.984 1	0.008 9	0.007 2	0.005 3
	Median Filter	0.999 4	0.002 2	0.997 0	0.003 7	0.983 5	0.007 2	0.007 4	0
	SG-Smoothing	0.998 6	0.003 5	0.992 9	0.005 6	0.969 0	0.041 0	0.015 1	0.038 3
	Normalize	0.997 9	0.009 4	0.989 7	0.006 8	0.988 3	0.006 4	0.006 3	0.001 6
	Baseline	0.998 3	0.003 9	0.992 0	0.006 0	0.986 3	0.006 7	0.006 7	-0.000 9
	SNV	0.999 0	0.002 9	0.995 6	0.004 4	0.992 8	0.004 9	0.004 9	-0.000 4
	MSC	0.999 1	0.002 7	0.996 3	0.004 1	0.991 8	0.005 4	0.005 3	-0.001 4

Table 3. MLR model based on optimal spectral band

View in the Article

Model	pre-treatment	R²	Name	Reference(W/W)	Predicted(W/W)	Deviation
MLR	SNV	0.992 8	CK1	0.000 5	0.002 6	0.004 4
			CK2	0.005 0	0.007 7	0.004 6
			CK3	0.015 0	0.014 2	0.004 3
			CK4	0.050 0	0.047 1	0.002 7
			CK5	0.110 0	0.111 4	0.003 0
			CK6	0.150 0	0.146 1	0.004 4
PLSR	Normalize	0.971 2	CK1	0.000 5	0.005 5	0.017 0
			CK2	0.005 0	-0.001 8	0.018 1
			CK3	0.015 0	0.018 5	0.017 3
			CK4	0.050 0	0.056 9	0.019 7
			CK5	0.110 0	0.102 7	0.025 8
			CK6	0.150 0	0.134 3	0.015 7
PCR	Raw	0.907 3	CK1	0.000 5	0.018 4	0.025 7
			CK2	0.005 0	0.010 9	0.019 4
			CK3	0.015 0	0.019 2	0.015 0
			CK4	0.050 0	0.059 6	0.015 4
			CK5	0.110 0	0.099 5	0.018 4
			CK6	0.150 0	0.125 2	0.011 8

Table 4. The predictive values of PLSR, PCR and MLR models compared with real value

Download Citation

Save the article for my favorites

Paper Information