Is is possible to have more detailed guidelines on pre/postprocessing?

[SoulflareRC]

Hi, I know this is a relatively old project, but I found several things that are unclear on preprocessing.

First, when I was reading the paper of this project (https://arxiv.org/pdf/1906.03720.pdf), I found that Figure 1 of the paper shows that preprocessing of the dataset happens before skull stripping, but its 3.1.1 Preprocessing says preprocessing("Adaptive window and level adjustment based on the image histogram to normalize intensities of tissues between cases, Z-score normalization of the entire data set") happens after skull stripping, so this confused me on when should the Matlab script be applied for preprocessing.

Second, there are two preprocessing.m in flair_segmentation and skull_stripping, and they are almost the same, so I'm wondering which script should I use to preprocess the data. Other than that, I copied the code for "Saving combined preprocessed slices from 3 modalities" from preprocessing.
m in flair_segmentation and applied the script on the dataset, but it's giving result that doesn't seem to be correct(see below).

Third, unrelated to preprocessing but I noticed that the skull stripping model sometimes tends to exclude the target area that is supposed to be segmented by the flair segmentation model due to the area's high "brightness", is there a way to solve this?