Yes, that can be done, Once the document is indexed, the index is usable without the original document being present.
The easiest way is to use the FILE_DATASTORE, so you're storing the path/filename of the file in the database, and the indexer pulls in the document text from the file system. Then, when your indexing is done, you can simply delete the document.
Beware, though, that rebuilding the index at some future time obviously won't be possible if you don't have the original document. Also you won't be able to use certain CTX_DOC services such as FILTER, HIGHLIGHT, MARKUP or SNIPPET which rely on accessing the original document.