The order of the images processed can vary by operating system, as each can have their own listing order quirks (case-sensitive or not, how they respond to spaces, etc). So I would highly suggest that you not rely on the order in which the data are processed, but instead to use the metadata from your filenames or elsewhere.
To do this, start with the Metadata module, which I guess you are not utilizing now? It can be daunting at first, but you can use this regular expression for your filenames, using "Extract from file/folder names" option:
^(?P<Row>.*) - (?P<Col>[0-9])\(fld (?P<Field>[0-9])
This will provide extra columns in your Per_Image table called Row, Col, Well (equal to concatenated RowCol), and Field.
Even better, you can use this extracted metadata in your ExportToDatabase module to calculate Per-well means, etc for all your measurements (though this does not work with ExportToSpreadsheet yet). But you could aggregate these yourself in R post-hoc much more safely, grouping by your plate/well metadata.
Hope that helps!