Table 1 Summary of the parameters, their values, their effects and their sensitivity on the segmentation accuracy

Image filtering

Contrast threshold (\({t}_{con}\))

60

60

Controls the degree of smoothening (i.e., noise removal) of the input image

Medium

Edge enhancement

Blur radius

1

0.9

Both parameters control the degree of sharpening of the image, i.e., increase the contrast at the boundaries between roots and pores and soil matrix

High

Mask weight

0.7

0.8

Background removal

Root gray value factor (\({f}_{r}\))

0.10

0.18

Sets the average gray value of the roots

Very high

Root gray value range (\({R}_{r}\))

65

70

Controls the root gray value window around the average root gray value. If set too high, overestimation of roots into their surroundings will occur. If set too low, loss of roots will occur

Very high

Detect fine roots

Minimum root diameter (\({d}_{r,min}\))

4

4

Controls the root recovery of the fine roots. If set too high, the fine roots are not detected. If set too low, over-segmentation may occur depending on the noise level of the image

High

Detect coarse roots

Maximum root diameter (\({d}_{r,max}\))

28

28

Controls the accuracy of the root diameter outline of the biggest root. If set too high, the diameter of the biggest root is overestimated and computational time is wasted. If set too low, the diameter of the biggest root is underestimated

Medium

False positives removal

Kernel size of median filter

3

2

Controls the degree of smoothening of the roots and trimming of over-segmented voxels. If set too high, root loss occurs whereas low values result in the presence of false positives

High

False negatives recovery

Size threshold (\({t}_{s}\))

25

25

Both parameters control the quality of the false negatives added in the root system. If set too low, many false positives are considered negatives, If set too high, root loss occurs

High

Vesselness threshold (\({t}_{v}\))

0.85

0.9

High