| 摘要: | 組織胺中毒為國際上常見的魚類食品中毒事件，而魚肉中組織胺的現有檢測方法，如液相層析、螢光光譜法等，皆需破壞樣品，且有較複雜的前處理步驟，不利於大規模且全面的檢驗。因此本研究的目的是應用近紅外光譜技術 (near infrared spectroscopy, NIRS) 結合多變量分析，開發出一種非破壞性快速檢測花腹鯖魚肉中組織胺的方法。本研究以類神經網路 (artificial neural network，ANN) 所建立之定性模型，對樣品中的組織胺含量是否符合臺灣所規範的限量標準 (≦200 ppm)進行判別，其正確率皆可達85%以上；而在對樣品是否達組織胺的毒性劑量 (>500 ppm) 進行判別時，也得到了不錯的結果。另以FOSS NIRS 6500型分光光度計測得新鮮花腹鯖魚片的光譜值，與液相層析串聯式質譜儀(liquid chromatography/tandem mass spectrometer, LC-MS/MS) 分析所得的組織胺數據，透過多重線性迴歸法 (multiple linear regression, MLR) 建立定量的校正模型。在低濃度樣品部分 (≦500 ppm)，結果顯示，魚肉側光譜模型的定量極限約為3 ppm，而表皮側光譜則為11 ppm；而在分析樣品中的組織胺含量是否符合臺灣所規範的限量標準 (≦200 ppm) 時，魚肉側光譜的檢量線適合範圍在9～200 ppm，而表皮側光譜則為 5～200 ppm；而在分析樣品中的組織胺含量是否達毒性劑量 (>500 ppm) 時，魚肉側與表皮側光譜的檢量線適合範圍皆在5～500 ppm。而高濃度樣品 (>500 ppm) 部分，以魚肉側光譜所建立的MLR模型效能最佳，其檢量線之相關係數為0.72， SECV為1012.41 ppm。本研究結果證實了近紅外光譜技術結合多變量分析法可以快速檢測花腹鯖魚肉中的組織胺含量。
Histamine fish poisoning, also called scombroid poisoning, is one of the most common types of food poisoning caused by fish consumption throughout the world. A variety of methods are available for determination of histamine in foods, including high performance liquid chromatography (HPLC) and spectrofluorometry. These methods, which are all destructive and require time-consuming sample preparation, are not suitable for large-scale inspection. The objective of this study was therefore to develop a rapid and non-destructive method for determination of histamine in blue mackerel (Scomber Australasicus) by using near infrared spectroscopy integrated with multivariate methods. In the present study, artificial neural network (ANN) was used to identify whether histamine concentrations in mackerel samples were higher than the limit (200 ppm) in Taiwan. The prediction accuracy of the ANN results were all above 85%. In addition, ANN models also showed good results for identification of the toxic level of histamine (500 ppm). The calibration models for quantification were built by multiple linear regression (MLR) through combining spectra of fresh mackerel fillets measured by FOSS NIRS 6500 spectrophotometer and histamine reference values determined by liquid chromatography/tandem mass spectrometer (LC-MS/MS). The limit of quantification (LOQ) of MLR calibration models built by the spectra of the flesh side of blue mackerel was 3 ppm, and which built by the spectra of the skin side was 11 ppm. Moreover, when determining whether histamine concentrations in mackerel samples were higher than the limit in Taiwan (200 ppm), the proper range of the calibration models built by the spectra of the flesh side of blue mackerel was from 9 to 200 ppm, while the skin side model ranged from 5 to 200 ppm. Further, when determining whether histamine concentrations exceed the toxic level (500 ppm), the proper ranges of the calibration models were 5 to 500 ppm in both flesh and skin side models. The MLR models showed the best result in second derivative spectra of the flesh side of blue mackerel which had histamine concentrations higher than 500 ppm, with RCV of 0.72 and SECV of 1012.41 ppm. In summary, integration of NIR spectroscopy and multivariate methods provides a useful tool for determination of histamine in blue mackerel (Scomber australasicus). |
