“…Image processing steps such as registration, filtering, segmentation, and quantification, are performed using a number of tools and software applications (refer to [ 13 , 22 , 27 , 32 , 61 , 67 , 68 ] for more examples). Some of the recent quantification methods combined with machine learning-based trainable algorithms for feature recognition (such as weka segmentation (e.g., [ 29 , 69 ]), and Seo et al, [ 70 ] are proven to be extremely helpful in working with noisy data to identify features on large data-sets ([ 71 ]). A single in situ tomographic experiment often results in few tens of 3D scans, and thus, robust image processing algorithms are crucial for processing a large number of data sets collected over several experimental cases.…”