Investigating the impact of spectral data pre-processing to assess honey botanical origin through Fourier transform infrared spectroscopy (FTIR)

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F60461373%3A22330%2F22%3A43925318" target="_blank" >RIV/60461373:22330/22:43925318 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Investigating the impact of spectral data pre-processing to assess honey botanical origin through Fourier transform infrared spectroscopy (FTIR)

Popis výsledku v původním jazyce

Honey is a valuable food commodity widely consumed around the world due to its unique organoleptic characteristics and significant nutritional value. Among the factors impacting honey market price, botanical origin plays a decisive role as certain types of honey are in a higher demand, e.g., honeydew honey [1]. To secure market sustainability and protect consumers from fraudulent acts, we developed a rapid and non-destructive Fourier transform infrared spectroscopy (FTIR) method successfully assessing honey botanical origin. To begin with, 39 honey samples originating from 9 botanical origins were collected during 3 seasons (2019, 2020, 2021). Upon measuring the samples absorbance in the mid-infrared region, a comprehensive evaluation of the spectral data pre-processing was performed, a practice commonly omitted or mispresented in the literature [2]. In detail, 16 different pre-processing methods were evaluated including both scatter-correction methods and spectral derivatives and their combinations. After developing partial least square discriminant analysis (PLS-DA) models for each pre-processing method, goodness of fit (R2Y) and goodness of prediction (Q2) were calculated. The dataset after second derivation by the GAP-segment algorithm (gap size - 3 points, segment size - 5 points) achieved the highest R2Y and Q2 values (R2Y= 0.885 and Q2=0.722) and among with 5 other pre-processing methods were further tested in a 5-classes sample set, specifically, blossom (n=15), thyme (n=7), cotton (n=5), fir (n=4) and orange (n=3) honeys. Again, the aforementioned GAP-segment algorithm obtained the best sample clustering (R2Y= 0.988 and Q2=0.880, for a 4-class model containing thyme, cotton, fir and orange honey samples) and was solely used to develop the final discriminatory models. Importantly, the blossom honey samples were excluded as the 5-classes model performance was rather unacceptable (R2Y= 0.263 and Q2= 0.063). This can be attributed to the composition of blossom honey containing nectar from multifloral sources. Six binary models were prepared for the four remaining classes and excellent sample clustering was performed with R2Y&gt;0.98 and Q2&gt;0.90 in every case. In conclusion, the presented study highlights the impact of data pre-processing strategies in vibrational spectroscopy and delivers a rapid and cost-efficient screening tool in honey authenticity testing.

Název v anglickém jazyce

Investigating the impact of spectral data pre-processing to assess honey botanical origin through Fourier transform infrared spectroscopy (FTIR)

Popis výsledku anglicky

Honey is a valuable food commodity widely consumed around the world due to its unique organoleptic characteristics and significant nutritional value. Among the factors impacting honey market price, botanical origin plays a decisive role as certain types of honey are in a higher demand, e.g., honeydew honey [1]. To secure market sustainability and protect consumers from fraudulent acts, we developed a rapid and non-destructive Fourier transform infrared spectroscopy (FTIR) method successfully assessing honey botanical origin. To begin with, 39 honey samples originating from 9 botanical origins were collected during 3 seasons (2019, 2020, 2021). Upon measuring the samples absorbance in the mid-infrared region, a comprehensive evaluation of the spectral data pre-processing was performed, a practice commonly omitted or mispresented in the literature [2]. In detail, 16 different pre-processing methods were evaluated including both scatter-correction methods and spectral derivatives and their combinations. After developing partial least square discriminant analysis (PLS-DA) models for each pre-processing method, goodness of fit (R2Y) and goodness of prediction (Q2) were calculated. The dataset after second derivation by the GAP-segment algorithm (gap size - 3 points, segment size - 5 points) achieved the highest R2Y and Q2 values (R2Y= 0.885 and Q2=0.722) and among with 5 other pre-processing methods were further tested in a 5-classes sample set, specifically, blossom (n=15), thyme (n=7), cotton (n=5), fir (n=4) and orange (n=3) honeys. Again, the aforementioned GAP-segment algorithm obtained the best sample clustering (R2Y= 0.988 and Q2=0.880, for a 4-class model containing thyme, cotton, fir and orange honey samples) and was solely used to develop the final discriminatory models. Importantly, the blossom honey samples were excluded as the 5-classes model performance was rather unacceptable (R2Y= 0.263 and Q2= 0.063). This can be attributed to the composition of blossom honey containing nectar from multifloral sources. Six binary models were prepared for the four remaining classes and excellent sample clustering was performed with R2Y&gt;0.98 and Q2&gt;0.90 in every case. In conclusion, the presented study highlights the impact of data pre-processing strategies in vibrational spectroscopy and delivers a rapid and cost-efficient screening tool in honey authenticity testing.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

10700 - Other natural sciences

Projekt

<a href="/cs/project/LM2018100" target="_blank" >LM2018100: Infrastruktura pro propagaci metrologie v potravinářství a výživě v České republice</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2022

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů