Towards automated detection of hyperhydricity in plant in vitro culture

Abstract Hyperhydricity (HH) is one of the most important physiological disorders that negatively affects various plant tissue culture techniques. The objective this study was to characterize optical features allow an automated detection HH. For purpose, HH induced in two species, apple and Arabidopsis thaliana , severity quantified based on visual scoring determination apoplastic liquid volume. comparison between score volume revealed a significant correlation, but different response dynamics. Corresponding leaf reflectance spectra were collected approaches spectral analyses evaluated for their ability identify HH-specific wavelengths. Statistical analysis raw showed significantly lower reflection hyperhydric leaves VIS, NIR SWIR region. Application continuum removal hull method identified absorption over time major peaks at 980 nm, 1150 1400 1520 1780 nm 1930 conducted experiments. Machine learning (ML) model spot checking specified support vector machine be suited classification explants, with test accuracy 85% outperforming traditional via vegetation index 63% other ML models tested. Investigations predictor importance 1950 1445 region 415 VIS classification. validity developed classifier tested available hyperspectral image acquisition SWIR-region.