AES and the Feistel Structure Used Together?

In Dual Watermarking in Tele-radiology using DWT for Data Authentication and Security, the authors wrote:

The encryption process uses the Feistel structure consisting multiple rounds for processing the plaintext to obtain the cipher text and each round consisting of a “substitution” followed by a permutation step. In this paper dual watermarking algorithm for patient detail encryption is used along with the key which uses a block cipher algorithm. Advanced Encryption Standard (AES) is a very powerful block cipher algorithm compared to the data encryption standard (DES) algorithm. It functions on a block of data at the input side with known size to execute same size of block of data at the output side. In this proposed system, the input blocks and output block data reserved is 128 bits of fixed length each. A symmetric key along with block cipher is taken where input to block cipher is AES. The AES encryption algorithm has the following stages of Byte substitution followed by row shifting followed by columns mixing and finally adding round keys.

Here, AES and the Feistel structure are both used. Please explain how this is possible with any kind of flow diagram.

Paper Published by K.Vinothini, S.K. Mydhili, S. Periyanayagi and G. Sukanya, "Dual Watermarking in Tele-radiology..."

Indeed, AES does not use the Feistel structure. The only way the quoted paragraph could make sense would be that "the encryption process" refers to prior work, distinct from the work in "this paper".

However, having skimmed through the paper (as on pages 0887 to 0891 of the proceedings), I fail to make sense out of it. The block diagram (reconstructed from the two in the paper) suggests some form of steganography or watermarking.

How the fingerprint used for encoding is similar enough to the one used for decoding in order to get meaningful results is left untold (the context seems to imply they are from different acquisitions). Except in the fingerprint subsystem, I do not see use of AES encryption.

Overall my impression is that the paper is written with the main (reached) goal of getting published (as countless other papers do), and only secondarily with the (failed) goal of trying to pass a cursory examination that it meaningfully describes something that was actually made to work. Stating nothing falsifiable about the system seems to be the main technique used towards the latter goal. It's not apparent that advancing the state of the art was in scope.

That lack of substance did not prevent the paper from getting cited, by this one, within three weeks of publication. This later paper also got cited.