Optimise parameters

[ellispatrick]

1. Run simpleSeg on multiple CODEX + IMC datasets to see if there is a set of parameters that work well on all of them.

2. Research metrics that could be used to optimise segmentation using a grid search. eg, signal to noise, roundness of nuclei....

[ellispatrick]

Predicting the segmentation mask with the marker that was used to predict it could be useful. This would allow us to extract a prediction probability and use this for the watershed, this might be more "standardised".

[ellispatrick]

Did Shila develop a package that could help with this?