Decision making base on the value of ghosting-signal-ratio

Hi all,
Does any one could share your experiences about how to use the value of the image quality metrics generated by MRIQC to rate if the data is good for use?
More specifically I want to hear you suggestion about the ghosting effect in the functional data. Since there is little we can do if we found ghosting effect in the data, I do want to understand how to decide if a data is still could be included in the analysis if there is a ghosting effect. Is there a certain threshold I can use for the gsr?
Thanks.