Extreme values
neurolabusc opened this issue · comments
Your project spurred a response from the AFNI team. One needs to be careful generating histograms to ensure your range and number of bins accurately captures the data. Thanks for bringing this interesting paper to my attention. We often neglect to discuss the handling of extreme values. It is clear that AFNI clips values that exceed z > 13.
Wow that's great! We found this issue a long time ago, and dug up that old paper with the fix and I implemented it here. I'm glad that it's been useful even after all these years! if you want to cite/reference the repository anywhere, there is a DOI associated with the repository:
https://zenodo.org/record/32508#.YTOsoFtMFH4
And of course Hughett's paper!
https://github.com/vsoch/TtoZ/blob/master/doc/JStats_Hughett.pdf
ping @jokedurnez and @chrisgorgo who were in my lab at the time and I think likely involved with working on this and would be interested to follow / participate in the discussion.
Thanks for the ping!
OK, I will close this issue as it was a FYI.