b score - how to handle empty wells

Hi all,

I'm hoping someone can help me with the Normalize Measurment B Score component.  I'm just looking at running it with Accumulate Statistics: For Each Plate and Include Controls: False.  Given what I see on the 2003 bscore paper, I think we should be using the raw value used for percent inhibition, but just the raw value.  However, I can't easily see how to handle are empty wells.  If I have a plate with just a couple odd empty wells (contamination or just empty) or a small bit of samples in a quadrant (imagine only one 96 well plate going into a 384) then is there a reasonable way to still create a reasonable bscore on the existing samples? I imagine filling empty wells with median values, mean sample values, or tagging them as controls are my options, but none of these work. 

Are there other options to attempt or a solid surrogate value someone has worked out for empty wells?

Many thanks,

Connie