pre-import naming of files as "lastname_firstname_servicedate_providercode_typeofdocument.pdf" did not avoid file losses and was instead associated with misfiles (misnamed, hence into the wrong patient file) so it was instead suggested that documents be simply
auto-named e.g. by date
auto-incremented e.g. number (PaperPort Visioneer, and Apple's Image Capture, do this)
import into the EMR was done visually, where the selected document was previewable (ideally) without having to be opened
Windows' Adobe will only provide readably large thumbnails for image files, not for PDFs
Apple's Preview application was valued for its ability to drag and drop pages from a left "source" document-and-pane into an adjacent new "target" file
naming convention for the document within the EMR was
type-of-study_body-region
supplemented by whatever key word or two (Dx in a report) that the clinician had highlighted (e.g. osteoarthritis or subcapital fracture)
worked equally well with consults
highlighting is much faster than writing and yellow highlighter doesn't show up in the scanned document
one practice found 150 dpi too degraded, so went with 300 DPI (others suggest 200 dpi)
5 years of paper scans, plus another several years' worth of old prior hand-written chart data, all scanned to PDF at 300 DPI black & white, added up to just about 12 GB in their backup, or about 1 Gb per full-time doctor-equivalent-year of general practice (contributed by a doctor on an Oscar McMaster list, Jan 15 2010)
make sure you don't inadvertently scan in colour or at too high a resolution; an auto-straighten (de-skew) function, where it is available, may increase file sizes
some scanning softwares or add-ons can also optically-character recognize (at accuracies of 97% or greater) and thereby adding a layer of searchable text beneath the scanned image