Advances in the acquisition and analysis of CT scan data to isolate a crop root system from the soil medium and quantify root system complexity in 3-D space

Root systems play an important role in soil-based resource acquisition. The ability to quantify the structural complexity of a root system should lead to an improved understanding of its many functions, but first of all, root systems must be visualized in the soil medium. Accordingly, the main objective of the reported study was to bring the novel approach based on computed tomography (CT) scanning to the next stage in root system studies, by upgrading the operational and analytical procedures available and reporting concrete results for a given crop species (i.e. maize) in different soil-moisture combinations. We also wanted to quantify the structural complexity of root systems by fractal analysis. Maize (Zea mays L.) was sown in standard plastic pots, 100 mm in top diameter. Sieved and homogenized mineral soils with no or very little organic material were used as growth media. A high-resolution X-ray CT scanner, formerly used for medical purposes, produced series of 500 CT images of 0.1-mm thick cross-sections of the root systems of the seedlings in dry and water-saturated soil conditions. Some seedlings were grown in a homogenous sandy soil, and were CT scanned about 5 days after emergence. Others were grown in loamy sand, and were CT scanned 3 days after emergence, because of their faster growth rate. Data analysis was performed on the 512 × 512 matrices of CT numbers used to construct CT images. Each CT number provided an indirect measure of density for one voxel with dimensions 0.12 × 0.12 × 0.1 mm3. We present the procedures that we used to digitally isolate a root system from the soil medium in 3-D space using CT scan data and to skeletonize the resulting 3-D image. We explain how to estimate the fractal dimension (FD) on such a skeletonized 3-D image of a root system in order to quantify its complexity. For the same root system, once removed from the soil and washed, FD was also estimated from a 2-D photograph. The expected dissimilarities between the FD estimates of the two types are confirmed and discussed. We found that the soil conditions allowing more accurate visualization of maize root systems by CT scanning were dry homogeneous sand and water-saturated loamy sand. The use of sieved and homogenized mineral soils facilitated the presentation of the new methodology here. In future studies, non-homogeneous soils can be used, and the guidelines that we have proposed will then be helpful in analyzing root geometry with respect to soil structure, for example.