I ran into this once while at the Medical College of Wisconsin
(someone had removed and inserted new media without unmounting
the first one). So everything ended up in the /lost+found directory.
Is that the case with you?

If so, the HEAD and BRIK files tend to be _near_ each other,
alphabetically, but it is not always the case. The timestamps
on the files may be helpful, but again, they are no reliably
useful. But if there are not too many datasets, you might be
able to recover some from that information.

Note that the HEAD files may contain the filename (as the
LABEL_1 attribute), and what may be the most useful is
the HISTORY_NOTE attribute, showing AFNI commands
used in creating the file.

If you find a potential pair, note that the BRICK_STATS
attribute might give you the range of values in the BRIK, so
you could verify whether that matches the raw BRIK data.

I hope some of this is helpful.

- rick