Determination of Sudan I in duck feed by microscopic image processing and confocal Raman spectroscopy

Duck feed is a globally traded agricultural product that is sometimes adulterated illegally with Sudan I for colour enhancement. This paper proposes a novel method combining microscopic image processing and confocal Raman spectroscopy to detect qualitatively the adulteration of duck feed. Image analysis of duck feed samples adulterated with Sudan I can rapidly and objectively locate highly suspicious areas to improve detection accuracy and speed. Confocal Raman spectroscopy is a valuable analytical tool in trace detection of Sudan I due to its high spatial resolution (below 1 mu m) and non-invasive nature. In this study, the microscopic images of Sudan I-spiked duck feed tablets (spiking levels: 0.0001-2%) are used to locate suspicious pixels that may contain the adulterant. The spectra of detected pixels were recorded using a confocal Raman spectrometer. An improved spectral angle measurement algorithm, feature-enhanced spectral angle measurement (FESAM), was used to enhance Raman spectral characteristics. Very good calibration statistic results were obtained to distinguish Sudan I from duck feed. In addition, model reliability was verified by other common additives that have similar colour to Sudan I. Results showed that an extension of the Sudan I adulterated calibration range increased the robustness of the model. FESAM effectively extracted and enhanced the characteristic peaks of low concentration Sudan I in duck feed, which decreased the limits of detection to 2.5 ppm. The whole detection process was completed very quickly (several minutes) without requiring sample pre-treatment. Therefore, Raman technology combined with image analysis data is a promising technique for detecting Sudan I adulteration in feed. The proposed method may be used to analyse other adulterated feeds with low concentrations of contaminants.