Virus infection assessment based on changes in spectral reflectance pattern on daughter seed potato plants

Fast and precise detection of infected potato plants is an essential practice in the seed potato certification system. Spectral fingerprinting as non-destructive and rapid method is going to be developed for discriminating plants with different stress such as disease. In this research virus infections of experimental plants (that infected with both PVY and PLRV viruses) were analyzed by spectral data without any destruction. Spectral data were collected from 32 plants (16 infected plants and 16 healthy plants) that were found to be infected or healthy using the ELISA and RT-PCR test. Some pretreatment methods of spectral data such as multiplicative scatter correction were used to remove noise. Soft independent modeling of class analogy (SIMCA) based on PCA analysis predicted the disease with high detection accuracy. The results showed, none of the samples belonged to the wrong group or to two groups simultaneously. The wavelengths in three ranges of 910-863 nm (near-infrared ), 725-704 nm (red edge) and 580-530 nm (green), had the greatest contribution to the complete differentiation of infected and healthy plants and development of models respectively.