Non-destructive Measurement of Color, Firmness and Lycopene Content of Tomato using Visible and NIR...
Transcript of Non-destructive Measurement of Color, Firmness and Lycopene Content of Tomato using Visible and NIR...
Non-destructive Measurement of Color, Firmness and Lycopene Content of Tomato using Visible and NIR Spectroscopy Limei ChenVijaya Raghavan, Ph.DDenis Charlebois, Ph.DMarie Thérèse Charles, Ph.D Clément Vigneault, Ph.D
50th Annual Conference/Conférence Annuelle13-16 July/juillet, 2008
Lonsdale Quay Hotel, 123 Carrie Cates CourtNorth Vancouver, British Columbia, Canada
Outline
Introduction Objective Materials and methods Results Conclusions
Tomato 2006 world production: 125 Tg*
Low in fat, calories and cholesterol-free; rich in vitamins A and C, lycopene, and β-carotene
Skin color and firmness - two important quality attributes for consumers**
References: *FAOSTAT online, http:// faostat.fao.org/default.aspx **Tijskens and Evelo, 1994. Postharvest Biology and Technology, 4, 85–89
Health benefits of lycopene A carotenoid imparting red color to f
ruits
Potent antioxidant: neutralize free radicals, protect body cells against oxidative damage
Possible preventive effect against several types of cancer (e.g. liver cancer, lung cancer and prostate cancer)
References: Clinton, 1998. Nutrition Reviews, 56, 35–51 Rao and Agarwal, 2000. Journal of the American College of Nutrition, 19, 563-569
Introduction of vis/NIR spectroscopyVis/NIR spectroscopy - a spectroscopic method utilising the visible and near infrared region of the electromagnetic spectrum (from about 380 to 2500 nm)
(Photo: http://www.baylor.edu/bucas/index.php?id=37025)
Principle of spectroscopy
NIR spectroscopy Large number of overtones and combination bands results in broad, poorly defined peaks
Higher signal-to-noise ratio compared to mid-IR spectroscopy
NIR application in food analysis
Good results for predicting SSC and dry matter
Grading line with NIR sensors are available from many companies (e.g. Aweta, Greefa, Mitsui-kinzoku)
Papers about fresh tomato analysis
Objective
Objective
Study the feasibility to evaluate the quality of tomatoes with non-destructive method based upon vis/NIR spectroscopy
Establish calibration models to predict color, firmness and lycopene content, simultaneously
Problem
Current methods to measure the lycopene content are laborious, destructive and organic solvents needed
Different instruments are employed for evaluation of tomato quality
Materials and methods 90 tomatoes Stored under 16ºC, 90-93% RH Spectra and quality parameters of 6 tomatoes measured
at 1, 5, 8, 12 and 16 days of ripening (DOR)
Cultivars 2 (cv. DRK 453 & cv. Trust )
Measuring times5
(breaker, turning, pink, light-red, red)
Tomatoes in each combination of
cultivar and DOR3
Collection times 3 (2 for calibration, 1 for validation)
Total No. of tomatoes = 2*5*3*3 = 90
Approach for building a spectroscopic method
Collecting spectral data (x-variables)
Determining quality parameters
(y-variables) by reference methods
Developing the calibration modelY = f (x)
Validating the model
Spectra acquisition
Spectrometer (FieldSpec® Pro FSP 350-2500P) coupled with a reflectance contact probe
6 (1 and 5 DOR) or 4 (8, 12 and 16 DOR) equidistant positions around the equator
Reflectance spectra 350-2500nm
Color and firmness measurement
Based on CIELAB system, color value L*, a*, b* measured by Minolta Chromameter
Firmness measured by a universal testing machine, expressed as peak force (N)
CIE L* a* b*
References: *Gómez et al., 2001. J. Sci. Food Agric, 81, 1101–1105 **Richardson and Hobson, 1987. J. Sci. Food Agric, 40, 245–252
(Photo: http://www.cigem.ca/pics/lab.jpg)
TCI = 2000a*/(L*(a*2+b*2)1/2)*
a*/b* ratio is a better index than a* in distinguishing varieties**
Lycopene content measurement
Homogenized and filtered to tomato juice extract
According to the reduced volumes of organic solvents method of Fish et al (2002), lycopene content determined by a spectrophotometer
Lycopene (mg/kg) = (A503 * 31.2) / (quantity of tissue used (g))
References: Fish et al., 2002. Journal of Food Composition and Analysis, 15, 309-317
Absorbance spectra560
nm
675
nm
1930 nm
980
nm
120
0
nm
1450 nm
Change of quality attributes
-0.4
0.0
0.4
0.8
1.2
1.6
0 2 4 6 8 10 12 14 16 18Day
Co
lor
a*/b
*5
10
15
20
25
30
Fir
mn
ess
(N)
color a*/b*
firmness
-20.0
-10.0
0.0
10.0
20.0
30.0
40.0
0 2 4 6 8 10 12 14 16 18
Day
TC
I
0
10
20
30
Lyc
op
ene
con
ten
t(m
g/k
g)
TCI
lycopene content
Multivariate calibration
Software package - “The Unscrambler v9.7”
Partial least squares (PLS) regression method was chosen due to the collinearity of x-variables
Full cross-validation used to determine the optimal PC number
Good PLS model: low RMSEC, RMSECV or RMSEP, high R2, SDR (SD/RMSEP)>3
Wavelength range selection
0.04
0.05
0.06
0.07
0.08
0.09
0.10
A B C D
RM
SE
CV
400
800
1,200
1,600
2,000
2,400
A B C D
Wavele
ng
th (
nm
)
For a*/b* 450-600 nm
Wavelength range selection
For TCI 430-1400 nm
2.00
2.20
2.40
2.60
A B C D
RM
SE
CV
400
800
1,200
1,600
2,000
2,400
A B C D
Wav
elen
gth
(n
m)
Wavelength range selection
For firmness 500-1100 nm
400
800
1,200
1,600
2,000
2,400
A B C D
Wav
elen
gth
(nm
)
2.00
3.00
4.00
5.00
A B C D
RM
SE
CV
(N
)
Wavelength range selection
2.00
2.50
3.00
3.50
4.00
A B C D E
RM
SE
CV
(m
g/kg
)
400
800
1,200
1,600
2,000
2,400
A B C D E
Wav
elen
gth
(nm
)
For lycopene content 450-1000 nm
Model resultsa*/b*
R2 = 0.99RMSECV=0.06SDR=8.89
-0.5
0
0.5
1
1.5
-0.5 0 0.5 1 1.5
Measured value
Pre
dic
ted
val
ue
Model results
R2=0.99 RMSECV=2.02SDR=8.84
TCI
Model resultsFirmness
R2 = 0.90RMSECV=2.24SDR=3.10
5
15
25
35
5 15 25 35
Measured value
Pre
dic
ted
val
ue
Model resultsLycopene content
R2 = 0.94RMSECV=2.57SDR=4.14
-5
5
15
25
35
45
-5 5 15 25 35 45
Measured value
Pre
dic
ted
val
ue
External validation
Parameters Wavelength range (nm)
Preprocessing method
No. of PCs
RMSEP R2 SDR
Color a*/b* 450-600 S.Golay 1st derivative
2 0.06 0.99 10.49
TCI 440-1400 - 3 1.52 0.99 11.74
Firmness 500-1100 - 4 1.44 0.97 6.12
Lycopene 450-1000 - 8 2.15 0.96 5.18
The result of firmness was superior to that reported by Shao et al. (2007) (r = 0.82, RMSEP=15.80) The result of lycopene was better than that reported by Baranska et al. (2006) using NIR (R2=0.85 and SECV=91.19)
Correlation of quality attributes
PLS2 model results
WavelengthNo. of
PCsParameters
Calibration Cross-validation External validation
RMSEC R2 RMSECV R2 SDR RMSEP R2 SDR
450-1100nm
6
Color a*/b* 0.05 0.99 0.05 0.99 10.57 0.06 0.99 8.89
TCI 2.25 0.98 2.49 0.98 7.17 1.75 0.99 11.65
Firmness 1.85 0.93 2.24 0.90 3.10 1.44 0.97 4.83
Lycopene 2.75 0.93 3.17 0.91 3.35 3.03 0.92 3.51
Conclusions It is feasible to build a fast and non-destructive
measurement of color, firmness and lycopene content of tomato fruits based on vis/NIR spectroscopy.
Calibration models with excellent performance were established to predict color value a*/b* ratio, tomato color index, firmness and lycopene content of tomato fruits.
Thank you