search

miykael / atlasreader

Python interface for generating coordinate tables and region labels from statistical MRI images
BSD 3-Clause "New" or "Revised" License
89 stars 31 forks source link

[FIX] probability threshold determines minimum proportion of a region of a cluster in clusters.csv #111

Closed AnnaD15 closed 4 years ago

AnnaD15 commented 4 years ago

Dear all,

first of all, thanks for this usefull tool! When I used the Harvard Oxford Atlas with a probability threshold of 25, I noticed that the probability threshold is not only used as a probability threshold for the atlas itself but the threshold also determines the minimum proportion of a cluster, which a region has to exceed to be part of the clusters.csv file. For instance a region encompassing 24% of a cluster would not be in the clusters.csv file after a probability threshold of 25 is applied. When changing the code according to the pull request, the region would be part of the clusters.csv file when a probility threshold of 0 or a threshold of 25 is applied. Only the proportion of the region would probably vary slightly, because of the different atlas probabilites used. In my mind the atlas probability threshold should not go in hand with the minimum proportion of a region in a certain cluster to be displayed in the clusters.csv file. Do you agree that the change makes sense? Or did I miss something?

Thank you Anna