Fig 1.
Influence of drifting grey values in individual scans on final combined volume grey value continuity.
(A) A single sample scanned in two parts (consistent scan parameters) with subtle compositional changes causing different grey values to be assigned. When digitally ‘stitched’ together, the resulting volume produces discontinuous grey values. (B) The same sample having applied the ‘Align Peaks’ method to produce uniform grey scale assignments across the two volumes. (C) Grey scale histograms of the individual top and bottom scans of [A] displaying the shift in grey value assignments and the relative similarity of the histograms. (D) The grey value histogram of the stitched images.
Fig 2.
Processing sequence of operations for segmentation of root system from CT scans.
For all images, the scale bar represents 10mm and the images depicted represent a small subsection of the total volume scanned (A) A single cross-section slice of the tomograph following image stitching of two sequential scans. The roots are centrally located and represented by mid grey values. (B) The same slice of the tomograph following resizing, where each voxel is expanded by a factor of two in each direction. (C) Single cross-section slice of tomograph following the identification and removal of partial volume effect pixels using a sobel filter. Isolated voxels are indicated in white tones. (D) Cross-section slice of tomograph after threshold is applied to isolate the root system. ‘Root’ voxels are represented by white. (E) Thresholded root system (white) following median and erosion filters to remove fine pore voxels connecting roots to non-root volumes. (F) 3D volume of the root system following selection using a connectivity algorithm.
Fig 3.
Example of segmentation output 4 vertically adjoining scans of chickpea (Cicer arietinum) var. Yorker root system 14 DAS.
Automated segmentation was used to extract the root system from CT scans using a single user determined seed point. Rendering performed using VG studiomax 2.0.
Fig 4.
Example of segmentation output 4 vertically adjoining scans of Wheat (Triticum aestivum L) var. Sunvale root system 28 days.
Automated segmentation was used to extract the root system from CT scans using a single user determined seed point. Rendering performed using VG studiomax 2.0. The raw root volume data was taken from the Flavel et al., 2014 data set.
Fig 5.
Example of segmentation of the 20 DAS barley root system.
A. High resolution (voxel size 27μm) and B. Low resolution (voxel size 57μm).
Fig 6.
Root length density (RLD) map of the of the wheat root system presented in Fig 3.
Cool colours represent low local RLD while warm colours indicate high levels of local RLD. High RLD in lower half of the volume are in response to a local phosphorus band. Rendering of 3D volume performed using VG Studio 2.0.
Fig 7.
Three dimensional visualisation of the root branching angle tool.
Root branch angle can be measured using thee points; along the parent root (1), at the branch vertex (2) and along the branch root (3). 3D root skeleton (red) is superimposed on the root segmentation (grey) and the angle calculated is indicated in green (broken line).
Fig 8.
ImageJ macros are accessible as a toolbar for easy access but can be readily customized for different experimental protocols.
The custom tools (identified from the default suite by the green color symbols) are (left to right) ‘Import tool’ for directly importing GE Sensing Technologies volume files into ImageJ; ‘Stitching tool’ which uses peak fitting to merge sequential scans for more consistent values; ‘rootone’ segmentation tool for extracting roots and export for connectivity filter; ‘Root Metrics’ tool which reduces volume size and measures skeleton characteristics; ‘Root length density map’ tool and root branch angle macro for manually measuring branch angle.