Determination of fat quality in pig meat and fat tissue using near-infrared spectroscopy

Preučevali smo točnost napovedovanja maščobnokislinske sestave maščobnega ter mišičnega tkiva z bližnjo infrardečo (NIR) spektroskopijo. Modele za napovedovanje smo razvili na osnovi spektrov vzorcev maščobnega in mišičnega tkiva ter referenčnih vrednosti kemijske analize za posamezne maščobne kisline (MK) in skupine MK (nasičene ̶ NMK, enkrat nenasičene ̶̶ ENMK, večkrat nasičene ̶ VNMK, n-3 in n-6 VNMK). Za napovedovanje smo uporabili laboratorijski NIR-spektrometer (model NIR System 6500) ter območje valovnih dolžin od 400 do 2500 nm. Vzorce maščobnega tkiva smo posneli v intaktni, vzorce mišic pa v homogenizirani obliki. Kakovost umeritvenih enačb smo ovrednotili s pomočjo parametrov standardna napaka kalibracije (seC), standardna napaka navzkrižne validacije (seCV), determinacijski koeficient kalibracije (R2C) in navzkrižne validacije (R2CV) ter parametrom RPD (razmerje med standardnim odklonom referenčnih vrednosti in seCV). Za zanesljive napovedi mora RPD presegati 3,0, vrednosti RPD med 2,0 in 3,0 pa kažejo na uporabnost enačb za grobo spremljanje. Rezultati kažejo precej uspešno napovedovanje skupin MK v podkožnem maščobnem tkivu z NIR-spektroskopijo. Najboljši rezultat smo dobili pri VNMK, kjer sta R2CV in RPD znašala 0,92 in 3,7, ter pri n-6 VNMK, kjer smo dosegli R2CV 0,92 in RPD 3,6. Za NMK sta R2CV in RPD znašala 0,76 in 2,0, pri ENMK 0,88 in 2,9 ter pri n-3 VNMK 0,66 in 1,7. Pri posameznih MK smo najboljše rezultate dobili pri vsoti C18:1 (R2CV = 0,85, RPD = 2,6), vendar je bil RPD pri večini MK pod 2,0. Pri skupinah MK so, z izjemo razmerja VNMK n-6/n-3, dobljene kalibracije uporabne za (grobo) spremljanje. Maščobnokislinske sestave mišičnega tkiva nismo uspeli zadovoljivo napovedati.The ability of near infrared (NIR) spectroscopy was tested for the prediction of fatty acid (FA) composition of fat and muscle tissue of pigs. Prediction models were developed based on the fat and muscle samples’ spectra and reference values, i.e. chemically analysed FA composition (individual FAs, saturated ̶̶ SFA, monounsaturated ̶̶ MUFA, polyunsaturated ̶̶ PUFA, n-3 and n-6 PUFA). A laboratory NIR spectrometer (model NIR System 6500) and a wavelength range from 400 to 2500 nm were used for the development of models. Fat tissue samples were scanned intact, while muscle samples were homogenized before scanning. The quality of models was assessed by means of standard error of calibration (seC) and cross-validation (seCV), coefficient of determination of calibration (R2C) and cross-validation (R2CV) and parameter RPD (the ratio between standard deviation of the reference values and seCV). For reliable predictions, the RPD must exceed 3.0, while values between 2.0 and 3.0 meet the requirements for rough screening. Our results showed successful prediction of FA groups in fat tissue by NIR spectroscopy. The best results were obtained for PUFA, where R2CV and RPD were 0.92 and 3.7, respectively and for n-6 PUFA, where R2CV of 0.92 and RPD of 3.6 were obtained. For SFA, R2CV and RPD were 0.76 and 2.0, for MUFA, 0.88 and 2.9, and for n-3 PUFA, 0.66 and 1.7, respectively. For individual FAs, the best results were obtained for the sum of C18:1 (R2CV = 0.85, RPD = 2.6), but for most FAs the RPD was below 2.0. The obtained models for FA groups are applicable for (rough) screening (with the exception of n-6/n-3 PUFA). The FA composition of the muscle tissue could not be adequately predicted