Fig. 7

Comparison of 3D and virtual 2D µCT (a–d) with 2D contact microradiography images (e–h) of Quercus faginea (Fagaceae) and the influence of different image plate resolutions and processing standards, all shown as X-ray negatives. a Penultimate step of the 3D data processing, in which non-target voxels, the epidermis and approximately 50% of the mesophyll tissue were removed during segmentation (modus: “Volume renderer Phong” with slight Erode-settings). The leaf venation was only eroded in the outermost layer but generally difficult to differentiate from surrounding parenchyma. This is due to the volume rendering procedure, which always represents rendering of surfaces. Thus, thickness information by density signals is lost. b Penultimate step of the 3D µCT image processing as a virtual 2D X-ray image (2D mode: “Sum along Ray” or—if more appropriate—“Maximum Projection”). The density values added up along a line of vision through the selected sub-volume showing that the highest vein orders are easier to discern from the surrounding parenchyma than in the 3D model due to particular density differences. c Final step of the 3D µCT image as a surface model (viewed from abaxial leaf surface). The skeletonization of the leaf venation is maximized by increasing the erosion with a slightly higher threshold. Part of the parenchyma is still visible and cannot be further faded out without impairing vein connectivity. d Final step of the 3D µCT image in virtual 2D X-ray mode. This step consists in the individually adjusted increase in contrast of the lower density values (dark tones) at the cost of a loss in contrast in the higher density values (light tones). The non-veinous mesophyll exhibits low density values (except for some granular structures) that are not confluent with the veins. Therefore, a high contrast of the leaf venation up to the highest vein orders is achieved. This processing mode is used subsequently as the reference µCT image. e Individual adjustment of the 2D X-ray image (20 µm) with better contrast and lower artificial widening (dilatation) of the veins compared to the standard protocol. f Individual adjustment of the 2D X-ray image (7 µm resolution). The increased resolution of the vein network compared to the industrial standard image is evident. Linear artefacts are visible which can be easily discerned from the veins. g Industrial standard (25 µm resolution) 2D X-ray image with simple contrasting according to our standard protocol. h Image processing of the 7 µm project standard (2D X-ray) according to our standard protocol. Leaf fragment 1 cm wide