Quantifying the effect of soil moisture content on segmenting root system architecture in X-ray computed tomography images

A commonly accepted challenge when visualising plant roots in X-ray micro Computed Tomography (mu CT) images is the similar X-ray attenuation of plant roots and soil phases. Soil moisture content remains a recognised, yet currently uncharacterised source of segmentation error. This work sought to quantify the effect of soil moisture content on the ability to segment roots from soil in mu CT images. Rice (Oryza sativa) plants grown in contrasting soils (loamy sand and clay loam) were mu CT scanned daily for nine days whilst drying from saturation. Root volumes were segmented from mu CT images and compared with volumes derived by root washing. At saturation the overlapping attenuation values of root material, water-filled soil pores and soil organic matter significantly hindered segmentation. However, in dry soil (ca. six days of drying post-saturation) the air-filled pores increased image noise adjacent to roots and impeded accurate visualisation of root material. The root volume was most accurately segmented at field capacity. Root volumes can be accurately segmented from mu CT images of undisturbed soil without compromising the growth requirements of the plant providing soil moisture content is kept at field capacity. We propose all future studies in this area should consider the error associated with scanning at different soil moisture contents.