A simple regex works fine, but transliteration is often more efficient. Here's an approach with tr///. This snippet iterates over each line of each file supplied at the command prompt, and prints the filename and line number, along with a highlighted version of the line's data showing where the problems are:

while( <> ) {
  chomp;
  if( tr/IMO//c ) {
    s/([^IMO]+)/[-->$1<--]/g;
    print "File: $ARGV, Line $.: $_\n";
  }
}
[download]

This would be invoked as:

It could be expressed as a one-liner, in fact, like this:

perl -lnE 'if(tr/IMO//c){s/([^IMO]+)/[-->$1<--]/g; say "$ARGV-$.: $_"}
+' file1.txt file2.txt fileXX.txt
[download]

The output is a little more terse, but it does the same thing. In either case, if your objective is just to print the bad lines, you can easily simplify the print statement.