search

miykael / atlasreader

Python interface for generating coordinate tables and region labels from statistical MRI images
BSD 3-Clause "New" or "Revised" License
88 stars 32 forks source link

Less clusters in csv file than output figures #145

Closed vigwma closed 3 months ago

vigwma commented 6 months ago

Thanks for this absolutely amazing toolbox!

I have used it successfully so far. Just one issue occured: For one contrast image the atlas reader detects 20 clusters. However the csv file shows only information for 9 clusters. Is that behavior expected or do I need to change the settings somewhere to display the region information for all cluster?

Thanks!

miykael commented 6 months ago

Hello, thanks for your question. I'm unfortunately not sure what's happening here. If it detects 20 clusters it should also report them. Only possible explanation (which it shouldn't) is that you chose some particular threshold or distance parameters that make it to not report the smallest clusters.

vigwma commented 6 months ago

Thanks for your reply. I chose cluster_extent to be 0 as well as voxel_thresh. For everything else I left the default values.

miykael commented 6 months ago

That might be the issue. I'm not sure how the algorithm reacts to having both of them 0. If you set them both slightly above 0 that might do the trick. So cluster_extent=1 (as it needs to be an integer, otherwise the parameter is used as a percentile) and voxel_thresh=0.1 (or something similar).

vigwma commented 6 months ago

Thanks for your fast reply! I did change the values but unfortunately it does not solve the issue...

Remi-Gau commented 6 months ago

@vigwma Do you think you could share a code example so we an reproduce the issue?

Ideally derived from our example code in this notebook? https://github.com/miykael/atlasreader/blob/master/notebooks/atlasreader.ipynb

vigwma commented 3 months ago

Hi @Remi-Gau, thank you so much for the response! I am very sorry for not replying so far. I had to work on a different part of the project. Now I was able to look at it again. I realized it was a file issue unrelated to the package. The package works perfectly now. Thank you so much!

Remi-Gau commented 3 months ago

ok I think this can be closed then @miykael