Median vs. mean roi values

Why are ROI values implemented as medians instead of means? Is there a reason to prefer one over the other for this application?

The calculation is over a numpy array with NaNs in it. At the time I wrote the module, nanmean wasn’t available, just nanmedian. Given that ROI arrays are usually normally distributed, it pretty close.

Makes sense, thanks for the response. The issue I’m having is that some of my CT data comes in as integers, and taking the median forces the result to be an integer. Not a major issue, but is there any chance of getting this changed in a future version?

As well, I’m not sure the assumption of normality is accurate, particularly for air on some CT scanners due to the range being bounded at -1000.


For backwards stability it will likely stay the same as the default. I will put in a ticket for a new setting.