Non-destructive evaluation of in vitro-stored plants:: A comparison of visual and image analysis

In vitro plants in slow-growth storage require routine evaluation for assessment of viability and need for repropagation. Determination of plantlet health by visual assessment is subjective and varies by genus due to variations in growth pattern and plant structure. Developing a standardized plant evaluation system would improve the efficiency of in vitro storage. This study was initiated to develop digital image analysis techniques for plantlets during slow-growth cold storage and to compare that system with visual examinations. Pear (Pyrus communis L.) cultivars were chosen for this initial trial because they have an open structure and clear internode position for image composition. Pear shoots stored at 4 degrees C in tissue culture bags were evaluated monthly by standard visual examination and by digital image analysis. Digital images were evaluated for red, green, blue, modified normalized differences of vegetation index (MNDVI), green/red ratio (G/R), intensity, hue, and saturation at the first two nodes of each plantlet. At 6 mo., the visual ratings had declined steadily for P. communis 'Luscious' and 'Bartlett-Swiss', while 'Belle Lucrative' and 'Louise Bonne de Jersey' ratings did not show significant declines until 9 mo. Correlations between visual ratings and G/R and MNDVI values were significant (r(2) >= 0.5) for all cultivars. Regression analysis indicated that the MNDVI and G/R ratios changed significantly over the 15-mo. rating period for most cultivars. Intensity, hue and saturation values were not consistently significant and did not correlate with visual ratings. These results will assist in the development of digital imaging as an alternative technique for evaluation of stored in vitro plantlets.