There is probably no correct answer to this question. All fixations will have some impact on the tissue content. It is usually necessary to simply be aware of the fact that the fixation protocol has not been completely benign and to just choose one protocol and then stick to it for all subsequent work. My experience to date has been completely consistent with the very good work reported in Chapter 3 of this book... http://accessengineeringlibrary.com/browse/vibrational-spectroscopic-imaging-for-biomedical-applications. We have also worked with freshly excised "wet" tissue.
What is the problem with the cryosections? Koljenovic has measured and published on dried cryosections of brain tissue without much problems. I also did measurements myself without much problems.
What is a problem if you would like to stain the dried sections afterwards. We circumvented this problem by staining the adjecent sections for pathology.
fresh samples were snap frozen in liquid nitrogen directly or Use isopentane to prevent fractioning. Cool the isopentane in liquid nitrogen until it becomes viscous. The samples were not embedded but merely glued with the OCT medium (at least, that is how I did it) onto the holders for cryo sectioning. This to prevent smearing the samples with the OCT medium. Samples were cut in 20 micron sections and dried in air. UV grade CaF2 glass and Fused Silica glass are both suitable as microscope slides. CaF2 has less signal but has more batch to batch variation, Fused Silica, as long as you use a single type, has a higher signal but has a very stable Raman signature that is not difficult to subtract.
For more details I suggest to approach Dr. S. Koljenovic directly, she is active on Research Gate as well.
Maybe not the most sophisticated protocol but it worked for us.