Fine tuning MultiTerm fuzzi-ness

I have a long TB I have built over time. I am using it in SDL Studio 2017 but it keeps suggesting/recognizing words which stray too much away from the term in the TB.

For example, I have in my TB the word placebo and when I run the QA in Studio it detects that the translation for placebo is missing in a segment. I look at the segment and it just contain the word place, which has nothing to do with it. It is okay a certain level of fuzziness, but I would like to limit it to the last letter (this is usually what changes between the male/female and singular/plural version of the same word in my language). I would also like that case matches the source and not the case in my TB.

 

For example if the word is written in upper case in the source and in my TB is in lower case, upon suggesting the term Studio should also suggest it in upper case. Is this possible, at all?