As a non-SmartBear person (I don't work for SmartBear), I can't really say what more there is you can do other than work with the indices. I haven't played around with ownerdrawn menus recently so I'm not sure if there is a way of getting the caption from other information. You might want to consider using what I mentioned above about TextRecognition to see if that will help you.
Other than readable code, is there any particular reason why you need to use the captions? Functionally, for the automation, it works the same.
