You could try blocking in donkey serum instead of BSA. Previously when I would do IHC or IF I used 5% goat serum in TBST as a blocking buffer when using secondary antibodies raised in goat. I have seen the percentages used vary between 3 and 10%. I used this buffer to dilute my primary and secondary antibodies as well.

There are several possibilities for what is going on:

1) autofluorescence. Compare against an unlabeled control (no primary or secondary antibody but washed and mounted the same way and imaged with the same settings). If this is the source, then you can do sodium borohydride washes to reduce the autofluoresence, or use a commercial autofluorescence blocking solution (such as the ReadyProbes™ Tissue Autofluorescence Quenching Kit).

2) Blocking isn't stringent enough. I would recommending adding 5% normal serum, or a commercial blocking solution such as BlockAid, and using it both before the primary antibody and in with the primary.

3) charge-based background binding. The Alexa Fluor dyes (and most dyes, for that matter) have a negative charge and will be attracted to positively charged areas on the tissue, such as myelin. You can use the Image-iT FX Signal Enhancer Solution at the beginning of your protocol to block the positive charges.

It is always best to use nonimmune serum from the animal that the second antibody was generated in. That way you are much more "specifically" blocking the non-specific binding of the secondary AB. Make up your primary AB in PBS containing 5% non-immune serum of the secondary AB source.