For a given project/collection we may want to identify a set of tags for an image. Tags could be important properties such as 'Stain', 'SubjectID', 'Region', etc. These fields can be random strings, but adding a basic validation schema is important when generating a searchable User Interface.
So for the case of a stain, I want to avoid having 'H&E', 'HE', 'H & E' as distinct stain types, as that makes it extremely difficult to find anything. So to create a validation scheme, for a given collection, I am creating a folder called .ProjectMetadata and in it, I am going to store any documents/PDF's, whatever that may describe the project itself.
Also, this folder is going to have a JSON object called # ProjectScheme that I use to keep track of the properties I want to validate and/or use for virtual folders.
As an example, I have .projectMetadata and
projectSchema { "validStains": [ "HE", "NISSL", "BISSL", "DISSL" ] }