CP2.2.0 np.histogram bug



In CellProfiler2.2.0 there seems to be a numpy related bug:

In case NaN values occur in an image this function crashes:

np.histogram(a, bins, range, normed, weights)


…/centrosome/threshold.py, line 815

The fix seems to be to explicitly exclude NaNs in the range argument of the np.histogram function.

Could you please fix it? We are still heavily relying on CP2.2.0 for analysis of our HT screens.


Hi Christian,

Can you upload this to GitHub, along with a pipelines and images to reproduce? Thanks!