Abstract- The objective of this study was to detect aflatoxin contamination in pistachio applying Raman spectroscopy technique and artificial neural networks. After spectra acquisition considering to principal components analysis(PCA)results, Second Derivative preprocessing method was selected and then principal components(PCs) were extracted to reduce the data dimensions. To classify samples, two kind of Multilayer perseptron feed forward back propagation topologies (including 8 neurons in hidden layer as first one and including 2 and 4 neurons in the first and second hidden layers respectively as a second one) were used. On average our classifiers performed with 92(one layer ANN) and 82(two layers ANN) percent accuracy . Both classifiers performance, for no contaminated samples recognition (as they achieved to 100 percent accuracy) was successful.

Keywords: Aflatoxin, ANN, PCA, Pistachio, Raman Spectroscopy

Full-Text [PDF 321 kb]   (1395 Downloads)