La spettroscopia dal campo alla tavola: composti bioattivi, autenticità

Transcription

La spettroscopia dal campo alla tavola: composti bioattivi, autenticità
Spettroscopia NIR e NIR-imaging per monitoraggio in situ e controllo
remoto in ambito agro-alimentare.
Milano, 24 settembre 2015
La spettroscopia dal campo alla tavola:
composti bioattivi, autenticità di prodotto e
monitoraggio di processo
Infrared spectroscopy from farm to fork: bioactive
compounds, product authenticity and process
monitoring
Ernestina Casiraghi
RATIONALE
INFRARED SPECTROSCOPY
Infrared spectroscopy has proven to be a successful analytical method for
quantitative and qualitative modelling of a wide variety of foodstuffs and
food production processes.
ADVANTAGES
• Speed of the analysis (10-60 s)
• Absence of chemical reagents
• Ease-of-use after initial method development
• Non destructive techniques
• Simultaneous measurements of several constituents
INFRARED SPECTROSCOPY and CHEMOMETRICS
Quantitative determination
PCA, Cluster analysis, LDA, KNN, PLS-DA, Class
modelling
Qualitative determination
MLR, PCR, PLS, Non-linear calibration
methods (ANN)
INFRARED SPECTROSCOPY and CHEMOMETRICS
Green and on-site tool
•to enhance food quality
•to confirm food authenticity
•to monitor food processing
IR and FOOD
number of pubblications
number of documents
600
Iran
France
Germany
Canada
Brazil
Italy
Spain
India
United States
China
500
400
300
200
100
0
2010
2011
2012
2013
2014
2015
2016
83
87
89
102
110
113
163
241
288
673
0
200
400
600
800
BIOACTIVE COMPOUNDS
BIOACTIVE COMPOUNDS: RATIONALE
AIM
Infrared spectroscopy to explore product
bioactivity
1. Development of a predictive methods for the
determination of phenolic compounds in grape skin
(by-product of wine industry) and blueberries;
2. Vitamin C content in Acerola by Hyperspectral Imaging.
Phenolic compounds in grape skin
High number of
samples from white
and red varieties on a
three years period
WINE INDUSTRY
BY-PRODUCTS
Barbera, Chardonnay (Vitis Vinifera L.)
GRAPE POMACES
Undistilled and distilled
Parameters
Method
Polyphenols
Folin-Ciocalteu assay
Anthocyanins
Abs 540 nm
Procyanidins
Acid-Butanol assay
Antioxidant
Activity
FRAP-Ferric Reducing
Antioxidant Power
Moisture
Oven at 105㼻C for 12 h
SKINS
SEEDS
RECOVERY
- Fibre fractions
- Sugar
- Phenolic compounds
APPLICATIONS
This research was supported by
AGER PROJECT
Grant n. 2010-2222
- Development of innovative functional foods
(baked goods)
- Natural antioxidants and antimicrobials
Phenolic compounds in grape skin
WET GRAPE SKINS
Partial least squares regression (PLS regression) models
Red grapes
White grapes
Dependent Variable
Min-Max
Data PreLV
Processing
Calibration
R2
RMSEC
Cross-Validation
R2
RMSECV
Moisture (g\100g)
2.56-11.45
SNV-d1
5
0.862 0.604
0.827
0.691
Total Polyphenols (GAEg/100g)
1.93-7.71
SNV-d1
7
0.893 0.376
0.771
0.567
Proanthocyanins (CYAEg/100g)
1.29-7.26
SNV-d1
7
0.800 0.614
0.700
0.755
Antioxidant Activity (TEmmol/100g)
9.89-43.34
SNV-d1
7
0.779 3.925
0.662
4.987
Moisture (g\100g)
2.43-10.08
SNV-d1
6
0.919 1.630
0.898
1.848
Total Polyphenols (GAEg/100g)
2.95-10.94
SNV-d1
7
0.850 1.025
0.814
1.156
Anthocyanins (CYAEg/100g)
0.08-2.93
SNV-d1
5
0.857 0.062
0.819
0.072
Proanthocyanins (CYAEg/100g)
1.96-12.92
SNV-d1
7
0.877 1.234
0.841
1.423
Antioxidant Activity (TEmmol/100g)
16.05-68.93
SNV-d1
7
0.905 14.091
0.864
17.064
SNV= Standard Normal Variate, d1= first derivative, LV= number of latent variables, R2= coefficient of determination,
RMSEC= root mean square error of calibration, RMSECV=root mean square error of cross-validation
Phenolic compounds in blueberries
Evaluation of quality and nutraceutical content of blueberries (Vaccinium
corymbosum L.) by near and mid-infrared spectroscopy. Postharvest Biology & Technology.
50, (1): 31-36, 2008. Sinelli, N., Spinardi, A., Di Egidio, V. Mignani, I., Casiraghi, E.
132 samples of blueberry (Vaccinium corymbosum L.)
PLS models to evaluate phenolic substances and vitamin C
Calibration
Dependent
variable
IR
method
LV
Total phenols
(mg cateching/g)
FT-NIR
10
FT-IR
Total flavonoids
(mg catechin/g)
Data
processing
Cross Validation
rcal
RMSEC
rcv
RMSECV
RPD
MSC
0.98
0.10
0.96
0.14
3.84
4
SNV
0.95
0.12
0.93
0.14
2.83
FT-NIR
7
SNV-d1
0.99
0.12
0.96
0.20
3.77
FT-IR
8
SNV
0.98
0.10
0.97
0.15
4.14
Total
anthocyanins
(mg malvidin/g)
FT-NIR
10
SNV
0.97
0.15
0.92
0.25
2.59
FT-IR
10
SNV
0.99
0.10
0.96
0.17
3.63
Ascorbic acid
(mg/100g)
FT-NIR
3
MSC
0.91
0.62
0.89
0.68
2.25
FT-IR
3
SNV
0.88
0.81
0.85
0.89
1.95
Vitamin C content in Acerola by
Hyperspectral Imaging
Acerola
(Malphighia emarginata)
v huge amount (from 1.0 to 4.5 mg/100 g) of
ascorbic acid
v green color, changing to yellow-reddish and
finally to purple
v a rapid deterioration is commonly observed,
due to its high moisture content
Vitamin C content in Acerola by
Hyperspectral Imaging
The AIM was the qualitative evaluation, by hyperspectral
imaging, of vitamin C distribution in acerola fruit as
discriminant quality parameter
RESULTS
The normalized coefficients of the two applied algorithms
allowed the visualization of the spatial distribution of pixels
highly correlated with vitamin C powder
HYPERSPECTRAL IMAGING FOR FOOD QUALITY
1. Acquisition of vitamin C pure spectrum
2. Evolution of juice spectra with added value of vitamin C powder
6700 – 7200
3. Selection of a region of the spectra (6700-7200 cm-1)
HYPERSPECTRAL IMAGING FOR FOOD QUALITY
4. Background removal using histogram of the intensity
background
fruit core
fruit
light
5. Application of SAM-spectral angle mapping and CC-correlation
coefficient on the hypercube of the selected region of the spectra
CORRELATION MAPS:
PRODUCT AUTHENTICITY
PRODUCT AUTHENTICITY: RATIONALE
AIM
Infrared spectroscopy to unravel
authenticity problems
1. Varietal discrimination of extra virgin olive oils by
near-and mid infrared spectroscopy
2. Fish species authenticity: Red mullet vs. Atlantic mullet.
Varietal discrimination of extra virgin olive oils
Varietal discrimination of extra virgin olive oils by near-and mid infrared
spectroscopy.
Sinelli N., Casale M., Di Egidio V., Oliveri P., Bassi D., Tura D., Casiraghi E. (2010)
Food Research International 43: 2126–2131.
82 monovarietal extra virgin olive oils
obtained by single-cultivars
C- CASALIVA (27 samples)
L- LECCINO (28 samples)
F- FRANTOIO (27 samples)
CHEMICAL DATA
PC3 (16%)
Principal component
analysis (PCA)
PC1 (25%)
Varietal discrimination of extra virgin olive oils
NIR
MIR
0.3
0.15
0.2
0.1
0.1
d2 [log(1/R)]
d2 [log(1/R)]
0.05
0
-0.1
-0.2
-0.3
12500
0
-0.05
-0.1
-0.15
-0.2
-0.25
10500
8500
6500
Wavenumbers
(cm-1)
SELECT ALGORITHM
(V-PARVUS package)
4500
3500
2800
2100
1400
Wavenumbers
(cm-1)
LINEAR DISCRIMINANT
ANALYSIS (LDA)
700
Varietal discrimination of extra virgin olive oils
LINEAR DISCRIMINAT ANALYSIS
(LDA)
NIR
MIR
Ø Class 1: Cv “Casaliva”
Ø Class 2: Cv “Leccino”
Ø Class 3: Cv “Frantoio”
Mean
Classification
ability (%)
Mean Prediction
ability (%)
Prediction
ability for
class 1 (%)
Prediction
ability for
class 2 (%)
Prediction
ability for
class 3 (%)
90.5
83.0
74.1
89.3
85.2
Mean
Classification
ability (%)
Mean Prediction
ability (%)
Prediction
ability for
class 1 (%)
Prediction
ability for
class 2 (%)
Prediction
ability for
class 3 (%)
94.2
86.6
81.5
93
85.2
The spectroscopic methods are able to discriminate the varietal origin of extra virgin
olive oils
PRODUCT AUTHENTICITY: RATIONALE
AIM
Infrared spectroscopy to unravel
authenticity problems
1. Varietal discrimination of extra virgin olive oils by near-and mid
infrared spectroscopy;
2. Fish species authenticity: Red mullet vs. Atlantic
mullet.
Red mullet vs. Atlantic mullet
NIR sphere- Diffusive reflectance
Range 12500-3750 cm-1 resolution
12cm-1; 64 scans
NIR fiber- Diffusive reflectance
Range 11000-4400 cm-1 resolution
12 cm-1;64 scans
IR - ATR
Range 4000-700 cm-1
resolution 4 cm-1
16 scans
132 RED MULLET–
Mullus surmuletus
165 ATLANTIC MULLET Pseudupeneus prayensis
Red mullet vs. Atlantic mullet
PCA results
FT-NIR sphere - smooth
FT-NIR fiber - SNV
䖃 Red Mullet
䖃 Atlantic Mullet
Good species distinction, especially in
the case of the optical fiber data
FT-IR - SNV+d1
Red mullet vs. Atlantic mullet
LDA results
Pre-treat.=smooth
RM (%)
AM (%)
mean (%)
External set 1
100
100
100
External set 2
100
100
100
External set 3
100
100
100
Pre-treat.=smooth
RM (%)
AM (%)
mean (%)
External set 1
100
100
100
External set 2
100
100
100
External set 3
100
100
100
AM (%)
mean (%)
Pre-treat.=SNV+d1 RM (%)
External set 1
100
100
100
External set 2
100
100
100
External set 3
100
100
100
All IR techniques
FT-NIR sphere
FT-NIR fiber
FT-IR
100% correct classification in prediction
PROCESS MONITORING
PROCESS MONITORING: RATIONALE
AIM
Infrared spectroscopy and Chemometrics to
study microbial food fermentations
1. Lactic acid fermentation in milk to describe curd
formation and, as a consequence, defining the end of
the fermentation;
2. Malolactic biotransformation in red wine: prediction of L-malic, Llactic and total acidity;
3. Wort fermentation to green beer: process trajectories and
regression for biomass, ethanol and solid content.
PROCESS MONITORING
To find a fast and reliable tool to get information about the main
modifications occurring during a food fermentation process
Infrared spectroscopy and Chemometrics
a successful analytical method if compared to conventional
analytical approaches
MCR-ALS
Multivariate Curve Resolution-Alternating
Least Squares
X
C
ST
E
It mathematically decomposes an instrumental
response for a mixture into the pure contributions
of each component involved in the kinetic of the
system studied.
LACTIC ACID (LA) FERMENTATION IN MILK
Monitoring of lactic acid fermentation process using Fourier Transform near
infrared spectroscopy.
Grassi S., Alamprese C., Bono V., Picozzi C., Foschino R., Casiraghi E. (2013)
Journal of NIR spectroscopy- Special Issue on Milk and Milk products, 21(4). DOI: 10.1255/jnirs.1058.
MONITOR TIME-RELATED CHANGES OCCURING IN LACTIC ACID FERMENTATION IN MILK:
PRINCIPAL COMPONENT ANALYSIS AND KINETIC MODELLING
Modelling Milk Lactic Acid Fermentation Using Multivariate Curve ResolutionAlternating Least Squares (MCR-ALS).
Grassi S., Alamprese C., Bono V., Casiraghi E., Amigo J.M.
Food and Bioprocess Technology. DOI: 10.1007/s11947-013-1189-2. In press.
MODELLING RHEOLOGICAL CHANGES IN MILK LACTIC ACID FERMENTATION BY MCR-ALS
Applicazione di tecniche spettroscopiche IR al monitoraggio della produzione di
latti fermentati
Alamprese C., Grassi S., Picozzi C., Bono V., Casiraghi E. (2012) 5th Italian Symposium of NIR Spectroscopy. NIR Italia
2012, Agripolis, Legnaro, Padua, September 26-28, 2012
DEVELOPMENT OF SPECIFIC CALIBRATION MODELS FOR TOTAL ACIDITY, AND MAIN SUGARS
INVOLVED IN THE FERMENTATION
LA FERMENTATION IN MILK: material & methods
S. thermophilus
Skim milk
(10% w/v)
37㼻C
S. thermophilus
+
L bulgaricus
41㼻C
L. bulgaricus
45㼻C
Monitored for
7h30min
• Dynamic oscillatory test (Physica MCR 300 rheometer) using concentric
cylinders and applying a constant 1% strain at a fixed 1 Hz frequency
• FT-NIR spectrometer (MPA, Bruker Optics) equipped with a
fiber-optic probe, 1 mm path length.
• Spectral range: 12,500-4,000 cm-1, resolution of 16 cm-1, 64 scans
Spectral range reduced to
8,900-5,555 cm-1
SNV
LA FERMENTATION IN MILK: methods
MCR-ALS
• Significant components: 3;
• Initial estimates: three spectra (beginning, middle and end of
fermentation batch);
• Constraints: non-negativity and unimodality to concentration
profiles;
• Convergence criterion: 0.1%.
Spectral
Concentration
NIR spectra (SNV)
Residuals
profile
profiles
E1
D1
C1
D2
C2
E2
D18
C18
E18
ST (F x N)
LA FERMENTATION IN MILK: results
MCR-ALS: concentration/status profiles
FASTER
F
A
S
T
E
R
99.9% of explained variance; %LOF: 0.63665; S.D. residuals <0.007
Influenced by
Liquid profile Transition
profile
Coagulated
milk profile
TEMPERATURE
INOCULUM
LA FERMENTATION IN MILK: results
Rheological Analysis
Dynamic oscillatory test: 1% strain, 1 Hz frequency
MALOLACTIC BIOTRANSFORMATION (MLF) IN WINE
Near and Mid Infrared Spectroscopy to detect malolactic biotransformation of
Oenococcus oeni in a wine-model.
Vigentini I., Grassi S., Sinelli N., Di Egidio V., Picozzi C., Foschino R., Casiraghi E.
Accepted in Journal of Agricultural Science and Technology
MONITOR TIME-RELATED CHANGES OCCURING IN MLF: PRINCIPAL COMPONENT
ANALYSIS AND KINETIC MODELLING
Near infrared and mid infrared spectroscopy in oenology: determination of main
components involved in malolactic transformation.
Grassi S., Vigentini I., Sinelli N., Foschino, R., Casiraghi, E.
NIR news (2012). doi:10.1255/nirn.1300.
DEVELOPMENT OF SPECIFIC CALIBRATION MODELS FOR MALIC ACID, LACTIC
ACID AND TOTAL ACIDITY
BEER FERMENTATION
Beer fermentation: Monitoring of process parameters by FT-NIR and
Multivariate data analysis
Grassi S., Amigo J.M, Bøge Lyndgaard,C., Foschino R., CasiraghiE.
Food Chemistry 155 (2014) 279–286
Beer fermentation monitoring by using FT-NIR spectroscopy
Grassi S., Amigo J.M., Bøge Lyndgaard C., Vigentini I., Casiraghi E. Proceedings in “NIR 2013 – 16th International
Conference on Near Infrared Spectroscopy” la Grande Motte, France, 2-7 June 2013
MONITOR TIME-RELATED CHANGES OCCURING IN BEER FERMENTATION : PRINCIPAL
COMPONENT ANALYSIS AND REGRESSION MODELS
Assesment of the sugars and ethanol development in beer fermentation with FT-IR
and Multivariate Curve Resolution models
Grassi S., Amigo J.M, Bøge Lyndgaard,C., Foschino R., Casiraghi E.
Submittedto Food Research International
MODELLING CHANGES IN SUGARS AND ETHANOL CONTENT IN WORT FERMENTING TO BEER
BY MCR-ALS
FUTURE PROSPECTIVES
Use of Hyperspectral
Imaging for bioactive
compounds detection
Development of NIR and
IR procedure for food
authentication purposes
Implementation of
infrared spectroscopy
sensors in food
industries
Acknowledgments and thanks
…To all the ones that contributed…
Dr.ssa Cristina Alamprese
Dr.ssa Veronica Bono
Dr.ssa Valentina Di Egidio
Prof. Roberto Foschino
Dr.ssa Silvia Grassi
Dr.ssa Cristina Malegori
Dr.ssa Claudia Picozzi
Dr.ssa Nicoletta Sinelli
Dr.ssa Ileana Vigentini
Ass. Prof. José Manuel Amigo
Dr. Christian Bøge Lyndgaard
Thanks!!