Limits...
Rapid phenotyping of crop root systems in undisturbed field soils using X-ray computed tomography.

Pfeifer J, Kirchgessner N, Colombi T, Walter A - Plant Methods (2015)

Bottom Line: Root systems from several crops were sampled in situ and CT-volumes determined with the presented method were compared to root dry matter of washed root samples.A highly significant (P < 0.01) and strong correlation (R(2) = 0.84) was found, demonstrating the value of the presented method in the context of field research.Application of the presented protocol helps to overcome the segmentation bottleneck and can be considered a step forward to high throughput root phenotyping facilitating appropriate sample sizes desired by science and breeding.

View Article: PubMed Central - PubMed

Affiliation: Institute of Agricultural Sciences, Swiss Federal Institute of Technology in Zurich (ETH Zürich), Universitätstrasse 2, 8092 Zurich, Switzerland.

ABSTRACT

Background: X-ray computed tomography (CT) has become a powerful tool for root phenotyping. Compared to rather classical, destructive methods, CT encompasses various advantages. In pot experiments the growth and development of the same individual root can be followed over time and in addition the unaltered configuration of the 3D root system architecture (RSA) interacting with a real field soil matrix can be studied. Yet, the throughput, which is essential for a more widespread application of CT for basic research or breeding programs, suffers from the bottleneck of rapid and standardized segmentation methods to extract root structures. Using available methods, root segmentation is done to a large extent manually, as it requires a lot of interactive parameter optimization and interpretation and therefore needs a lot of time.

Results: Based on commercially available software, this paper presents a protocol that is faster, more standardized and more versatile compared to existing segmentation methods, particularly if used to analyse field samples collected in situ. To the knowledge of the authors this is the first study approaching to develop a comprehensive segmentation method suitable for comparatively large columns sampled in situ which contain complex, not necessarily connected root systems from multiple plants grown in undisturbed field soil. Root systems from several crops were sampled in situ and CT-volumes determined with the presented method were compared to root dry matter of washed root samples. A highly significant (P < 0.01) and strong correlation (R(2) = 0.84) was found, demonstrating the value of the presented method in the context of field research. Subsequent to segmentation, a method for the measurement of root thickness distribution has been used. Root thickness is a central RSA trait for various physiological research questions such as root growth in compacted soil or under oxygen deficient soil conditions, but hardly assessable in high throughput until today, due to a lack of available protocols.

Conclusions: Application of the presented protocol helps to overcome the segmentation bottleneck and can be considered a step forward to high throughput root phenotyping facilitating appropriate sample sizes desired by science and breeding.

No MeSH data available.


Related in: MedlinePlus

Steps of the segmentation protocol. Original X-ray CT volume of grass-legume mixture sample (details given in Table 1) showing roots, air-filled pores and soil (a). First step: advanced surface determination of the soil. The surface is shown as a blue line around the soil aggregates (b). Second step: dilatation of the region of interest (ROI), here 1 voxel, to add mixed voxels. The contour of the dilated surface is shown as a bright blue line (c). Step three: subtraction of the dilated ROI from a ROI containing the whole volume. Only roots and pores remain in the resulting volume (d). Step four: detection of the root surface (shown as a blue line) (e). Step five: the volume containing the roots and remaining noise (f) is exported to MatLab and filtered therein. The resulting, filtered volume containing only roots is shown in (g). The peaks of the gray values of air, mixed voxels, roots and minerals shown in the histogram are not completely separated (h)
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4552158&req=5

Fig2: Steps of the segmentation protocol. Original X-ray CT volume of grass-legume mixture sample (details given in Table 1) showing roots, air-filled pores and soil (a). First step: advanced surface determination of the soil. The surface is shown as a blue line around the soil aggregates (b). Second step: dilatation of the region of interest (ROI), here 1 voxel, to add mixed voxels. The contour of the dilated surface is shown as a bright blue line (c). Step three: subtraction of the dilated ROI from a ROI containing the whole volume. Only roots and pores remain in the resulting volume (d). Step four: detection of the root surface (shown as a blue line) (e). Step five: the volume containing the roots and remaining noise (f) is exported to MatLab and filtered therein. The resulting, filtered volume containing only roots is shown in (g). The peaks of the gray values of air, mixed voxels, roots and minerals shown in the histogram are not completely separated (h)

Mentions: Volume data analysis was performed by VG Studio MAX 2.2 software (Volume Graphics GmbH, Heidelberg, Germany) and the add-on modules ‘Coordinate measurement’ (Advanced surface determination) and ‘Wall thickness analysis’. Original images (32-bit float) were downscaled to 16-bit unsigned integer. In general, pores and air are of lower gray values than roots, which are of lower gray values than soil components. Unfortunately, the four mentioned objects are not easily separable by simple global thresholding [9, 10]. In the first step of this protocol, all mineral structures (soil) were segmented in the original reconstructed volume (Fig. 2a) using the ‘Advanced surface determination’-tool by manual selection of air as background and mineral parts as material using the ‘Define material by example area’-function (Fig. 2b). Applying the ‘Advanced surface determination’-tool, gray value thresholds can be continuously adjusted according to a preview window showing the resulting surface determination. The ‘Advanced surface determination’-tool refines the surface locally at several thousand locations along the object surface (here: soil aggregates) by a local adjustment according to the gradient of the gray values. The same gray value is reinterpreted according to the gray value of the neighboring voxels. This allows for a very precise determination of the surface to the target structure. A new region of interest (ROI) was generated from the surface, and attention was paid that preferably no root voxels but, where possible, all mineral voxels were included. In the second step (Fig. 2c), the ROI was dilated by 0.5–1 voxels in order to add mixed voxels at the border of the soil aggregates to air-filled pore spaces (also missing mineral voxels are commonly added by dilatation). Mixed voxels are formed due to volume averaging effects and frequently have gray values similar to root voxels, which would hinder root segmentation significantly [10]. In step three (Fig. 2d), the ROI containing mineral structures and mixed voxels was then subtracted from a ROI containing the whole volume so that only roots and pores filled with air and water remain in the resulting volume (Fig. 2d). Here, it proved very helpful to choose pots made of PVC as the gray value of PVC is very different to the one of roots and normally similar to the gray value of the mineral fraction. For this reason, the pot wall is included in the ROI containing mineral structures. In step four (Fig. 2e), the roots were segmented within the volume containing only roots and pore spaces analogously by manual selection of air as background and roots as material using the ‘Define material by example area’-function. A volume containing all roots and some remaining noise can be generated by extracting a new volume from the root surface. The fifth step comprises noise elimination, which was performed on exported tiff image stacks (Fig. 2f) by MatLab 8.0 (The Mathworks, Natick, Massachusetts, United States) using size thresholds. Using the MatLab algorithm (script provided as an.exe-file in Additional file 1, working on 16-bit unsigned integer) all structures smaller than 10,000 connected voxels were deleted. The denoised volumes were saved as tiff image stacks and imported to VG Studio MAX 2.2 for subsequent analysis (Fig. 2g).Fig. 2


Rapid phenotyping of crop root systems in undisturbed field soils using X-ray computed tomography.

Pfeifer J, Kirchgessner N, Colombi T, Walter A - Plant Methods (2015)

Steps of the segmentation protocol. Original X-ray CT volume of grass-legume mixture sample (details given in Table 1) showing roots, air-filled pores and soil (a). First step: advanced surface determination of the soil. The surface is shown as a blue line around the soil aggregates (b). Second step: dilatation of the region of interest (ROI), here 1 voxel, to add mixed voxels. The contour of the dilated surface is shown as a bright blue line (c). Step three: subtraction of the dilated ROI from a ROI containing the whole volume. Only roots and pores remain in the resulting volume (d). Step four: detection of the root surface (shown as a blue line) (e). Step five: the volume containing the roots and remaining noise (f) is exported to MatLab and filtered therein. The resulting, filtered volume containing only roots is shown in (g). The peaks of the gray values of air, mixed voxels, roots and minerals shown in the histogram are not completely separated (h)
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4552158&req=5

Fig2: Steps of the segmentation protocol. Original X-ray CT volume of grass-legume mixture sample (details given in Table 1) showing roots, air-filled pores and soil (a). First step: advanced surface determination of the soil. The surface is shown as a blue line around the soil aggregates (b). Second step: dilatation of the region of interest (ROI), here 1 voxel, to add mixed voxels. The contour of the dilated surface is shown as a bright blue line (c). Step three: subtraction of the dilated ROI from a ROI containing the whole volume. Only roots and pores remain in the resulting volume (d). Step four: detection of the root surface (shown as a blue line) (e). Step five: the volume containing the roots and remaining noise (f) is exported to MatLab and filtered therein. The resulting, filtered volume containing only roots is shown in (g). The peaks of the gray values of air, mixed voxels, roots and minerals shown in the histogram are not completely separated (h)
Mentions: Volume data analysis was performed by VG Studio MAX 2.2 software (Volume Graphics GmbH, Heidelberg, Germany) and the add-on modules ‘Coordinate measurement’ (Advanced surface determination) and ‘Wall thickness analysis’. Original images (32-bit float) were downscaled to 16-bit unsigned integer. In general, pores and air are of lower gray values than roots, which are of lower gray values than soil components. Unfortunately, the four mentioned objects are not easily separable by simple global thresholding [9, 10]. In the first step of this protocol, all mineral structures (soil) were segmented in the original reconstructed volume (Fig. 2a) using the ‘Advanced surface determination’-tool by manual selection of air as background and mineral parts as material using the ‘Define material by example area’-function (Fig. 2b). Applying the ‘Advanced surface determination’-tool, gray value thresholds can be continuously adjusted according to a preview window showing the resulting surface determination. The ‘Advanced surface determination’-tool refines the surface locally at several thousand locations along the object surface (here: soil aggregates) by a local adjustment according to the gradient of the gray values. The same gray value is reinterpreted according to the gray value of the neighboring voxels. This allows for a very precise determination of the surface to the target structure. A new region of interest (ROI) was generated from the surface, and attention was paid that preferably no root voxels but, where possible, all mineral voxels were included. In the second step (Fig. 2c), the ROI was dilated by 0.5–1 voxels in order to add mixed voxels at the border of the soil aggregates to air-filled pore spaces (also missing mineral voxels are commonly added by dilatation). Mixed voxels are formed due to volume averaging effects and frequently have gray values similar to root voxels, which would hinder root segmentation significantly [10]. In step three (Fig. 2d), the ROI containing mineral structures and mixed voxels was then subtracted from a ROI containing the whole volume so that only roots and pores filled with air and water remain in the resulting volume (Fig. 2d). Here, it proved very helpful to choose pots made of PVC as the gray value of PVC is very different to the one of roots and normally similar to the gray value of the mineral fraction. For this reason, the pot wall is included in the ROI containing mineral structures. In step four (Fig. 2e), the roots were segmented within the volume containing only roots and pore spaces analogously by manual selection of air as background and roots as material using the ‘Define material by example area’-function. A volume containing all roots and some remaining noise can be generated by extracting a new volume from the root surface. The fifth step comprises noise elimination, which was performed on exported tiff image stacks (Fig. 2f) by MatLab 8.0 (The Mathworks, Natick, Massachusetts, United States) using size thresholds. Using the MatLab algorithm (script provided as an.exe-file in Additional file 1, working on 16-bit unsigned integer) all structures smaller than 10,000 connected voxels were deleted. The denoised volumes were saved as tiff image stacks and imported to VG Studio MAX 2.2 for subsequent analysis (Fig. 2g).Fig. 2

Bottom Line: Root systems from several crops were sampled in situ and CT-volumes determined with the presented method were compared to root dry matter of washed root samples.A highly significant (P < 0.01) and strong correlation (R(2) = 0.84) was found, demonstrating the value of the presented method in the context of field research.Application of the presented protocol helps to overcome the segmentation bottleneck and can be considered a step forward to high throughput root phenotyping facilitating appropriate sample sizes desired by science and breeding.

View Article: PubMed Central - PubMed

Affiliation: Institute of Agricultural Sciences, Swiss Federal Institute of Technology in Zurich (ETH Zürich), Universitätstrasse 2, 8092 Zurich, Switzerland.

ABSTRACT

Background: X-ray computed tomography (CT) has become a powerful tool for root phenotyping. Compared to rather classical, destructive methods, CT encompasses various advantages. In pot experiments the growth and development of the same individual root can be followed over time and in addition the unaltered configuration of the 3D root system architecture (RSA) interacting with a real field soil matrix can be studied. Yet, the throughput, which is essential for a more widespread application of CT for basic research or breeding programs, suffers from the bottleneck of rapid and standardized segmentation methods to extract root structures. Using available methods, root segmentation is done to a large extent manually, as it requires a lot of interactive parameter optimization and interpretation and therefore needs a lot of time.

Results: Based on commercially available software, this paper presents a protocol that is faster, more standardized and more versatile compared to existing segmentation methods, particularly if used to analyse field samples collected in situ. To the knowledge of the authors this is the first study approaching to develop a comprehensive segmentation method suitable for comparatively large columns sampled in situ which contain complex, not necessarily connected root systems from multiple plants grown in undisturbed field soil. Root systems from several crops were sampled in situ and CT-volumes determined with the presented method were compared to root dry matter of washed root samples. A highly significant (P < 0.01) and strong correlation (R(2) = 0.84) was found, demonstrating the value of the presented method in the context of field research. Subsequent to segmentation, a method for the measurement of root thickness distribution has been used. Root thickness is a central RSA trait for various physiological research questions such as root growth in compacted soil or under oxygen deficient soil conditions, but hardly assessable in high throughput until today, due to a lack of available protocols.

Conclusions: Application of the presented protocol helps to overcome the segmentation bottleneck and can be considered a step forward to high throughput root phenotyping facilitating appropriate sample sizes desired by science and breeding.

No MeSH data available.


Related in: MedlinePlus