Semi-automated segmentation and visualization of complex undisturbed root systems with X-ray μCT

Abstract

Increasing interest in plant-root phenotyping has stimulated the development of X-ray μCT-based root/soil segmentation protocols. However, most scanning and CT volume processing protocols were only applied for detection of simple and small juvenile roots. We tested a new methodology for its ability to extract large mature maize roots from X-ray μCT volumes of undisturbed soil monoliths (10 × 20 × 15 cm) and compared its performance with two existing (semi-)automated segmentation algorithms. The X-ray μCT based root assessments were validated by means of the measured root biomass. Segmentation of maize root systems proved to be particularly challenging because regardless of soil type, within-sample vertical gradients in X-ray attenuation caused an overlap in CT-reconstructed grey values of roots, water and mineral phases. However, in contrast to the two existing algorithms, a new alternative methodology did allow for relatively fast and accurate segmentation of the mature roots. This was evidenced by a highly significant correlation between the CT-derived root volume and its measured root biomass. Despite the large size of the investigated soil monoliths, roots could be resolved down to a diameter of 200 μm. This methodology opens up new possibilities for investigating large and complex root systems, and may become an invaluable new tool in plant breeding research.