Starting from what appears to be your main question:
Can I use SNPs associated with a gene's higher expression to compute the likelihood of that gene being expressed in the brain region?
I would strongly advise against using SNPs determine if genes are expressed (at all) in a given tissue.For one thing SNPs that affect expression (then often called eQTLs) will usually only have small +- effects on expression instead of switching it on/off and also have that effect in many/all cell types. Unless an (individual) SNP has clearly described /and validated effects on gene expression in a specific tissue/brain region, you should assume it has global effects.
Now, luckily for you gene expression in different cell types, tissues etc is a relatively well studied area and there are resources for gene expression in the brain available (brain atlas is probably closest to what you might look for).
You can uses these resources to determine which genes are expressed in your regions of interest and from there on check if SNPs associated with these genes correlate with certain structural brain features you are interest in.
One more thing you note:
The title of your questions (as well es comments) mention that you are interested in epigenetics. Neither gene expression nor SNPs/eQTLS are diretcly linked to epigenetics, so if this is your actual focus you should also look for availbale ChIP-seq data or similar and also include this into your analysis (correlation of SNPs with epigenetic marks is difficult but not impossible)