Digital watermarking is a mechanism for protecting digital documents as a complement to cryptography. Despite the many solutions proposed in the literature, to our knowledge there is no universal solution that can be applied everywhere. In this paper, we propose a new hybrid medical image watermarking approach that is based on two watermarks. The first watermark (the image signature) is based on three static parameters (entropy, Hu moment and co-occurrence matrix) of the image and a chaotic iteration. The second watermark is based on the image identifier. This identifier is based on the patient identifier and image information. The patient identifier is generated from the patient information. This second watermark, which is unique, must be inserted into one of the image coefficients. Several techniques have been associated with the insertion of the watermark and must guarantee good quality of the watermarked image. The simulation results show that our approach can generate unique identifiers for the image and the patient with very few parameters. We also have a good quality image watermarked. The system will detect any attempt to modify the image or patient information. In our approach, the watermarked image is encrypted using an asynchronous encryption algorithm to reconstitute the image after an attack.  Image compression has also been taken into account.