Suggestions for improvement (Setup and first lesson, NIfTI)

[RhiannonJeans]

First I would like to say I have been enjoying the lessons so far. I have completed Setup and Anatomy of a NIfTI. It is my first introduction to looking at fMRI data in Python (or any programming language), so I appreciate how these learning resources have been made so accessible and approachable to a newbie. I hope I can provide some helpful comments from a beginner's perspective. I also have limited experience with git, so this was my first time downloading a git archive.

First, some comments on the setup page.

• Laid out very clearly in terms of formatting, dependencies, required installations
• I personally found the conda install instructions for dcm2niix most useful, perhaps these instructions could be in Setup.
• I could not get DataLad to work, and it is not immediately clear to me what benefit DataLad has over downloading the git. In addition, I did not understand the path '///openneuro/ds000030'. It's not obvious to me what the root folder is before ///, and if this is specific to unix or something else. However I found a solution to this.
• I ended up downloading the materials using 'git clone https://github.com/carpentries-incubator/SDC-BIDS-IntroMRI.git' and put it somewhere on my C:/Users/[user]/neuroimaging as git had an issue with the default directory lacking space (not sure why this is, perhaps irrelevant).
• Regarding how to load the data. I intially tried Binder. Turning off ads seemed to help the main binder directory load but none of the files seemed to be linked as they lead to an error when loading. For this reason I ended up going with the Jupyter Notebook option on Windows 10 and installing any packages via Anaconda Navigator or using the terminal when they were available via pip or conda only.

Next, some brief comments about lesson 1, Anatomy of a NIfTI

• The instructions in this tutorial were very easy to follow, were informative, and I encountered very few errors.
• The instructions "type t1_img. followed by Tab" does nothing on my machine or Jupyter Notebook setup. I am using a custom theme, monokai, so I cannot be sure if this is the issue or if this is specific to another platform. (Update: The solution is to update the jedi package).
• In one place "<kbdTab" it is not formatted correctly in the document.
• When specifying x_slice, I think it should be [9,:,:] not [9:,:,:]
• One other minor formatting issue " ((\text{Right},\text{Anterior},\text{Superior}))"

I hope this helps!

[josephmje]

@RhiannonJeans Thanks so much for working through this lesson and the detailed feedback! This is really helpful! I will work on including these suggestions soon.

I personally found the conda install instructions for dcm2niix most useful, perhaps these instructions could be in Setup.

Good point! Especially because the lesson starts off by recommending conda for installing Python

I could not get DataLad to work

DataLad is a similar tool to git but it allows you to version control much larger files like MRI data. The actual lesson where DataLad is being used is still in the works. The Jupyter notebook is present in the code folder but hasn't been converted to the Carpentries template yet. I think the setup would benefit from more description about what DataLad is.

I did not understand the path '///openneuro/ds000030'. It's not obvious to me what the root folder is before ///

The /// is a DataLad resource identifier. The command is attempting to download the dataset ds000030 from openneuro.org. We should explain this better.

Turning off ads seemed to help the main binder directory load but none of the files seemed to be linked as they lead to an error when loading.

Thanks for bringing this up. I'll need to check on this.

The instructions "type t1_img. followed by Tab" does nothing on my machine or Jupyter Notebook setup.

Thank you for finding a solution to this. I hadn't encountered this before but good to know that we only have to update the jedi package.

Will update all of those formatting issues you noticed. We recently added automated website build previews. Any changes to the lesson will create a temporary website to make sure things render properly. This will hopefully help us avoid those formatting issues going forward.

[RhiannonJeans]

Happy to help! To clarify:

Turning off ads seemed to help the main binder directory load but none of the files seemed to be linked as they lead to an error when loading.

What I mean by this is when I turn off ad blocker it enabled the binder directory to load, but not the files. I think I may have accidentally implied keeping the ad blocker on was causing it.

The /// is a DataLad resource identifier. The command is attempting to download the dataset ds000030 from openneuro.org. We should explain this better.

If it helps, I tried to install DataLad with conda as well. I can't be sure if this is the cause of the error. In any case I think more details about setting up DataLad and loading the dataset would be beneficial.

[YuyangXueEd]

Thank you very much for your lesson. I am new to MR Imaging and I learned a lot from this tutorial. However, there are something I don't understand and some errors I found in some of courses:

1. In the lesson of 2. Anatomy of a NIfTI in carpentries-incubator.io, it is different from the notebook version of 2. Anatomy of a NIfTI. There is a missing coordinate in the context of "Suppose we have two voxels located at the the following coordinates ... ". And the result of two affined corrdinate is also missing one of them.
2. I have used the dataset and code provided by the notebook. The t1_affine gives the affine matrix which is different from the tutorial. Would you mind complement the process of how the coordinates are manipulated by the affine matrix? It would be much help.
3. I have also write a more "Pythonic" version of Setup notebook and Scanner_to_computer notebook in my own repo. Maybe it could do some help to people who use Windows OS or other distributions which not support bash interpreter :)

Thanks again for all your effort 🥇