Reversible Data Hiding Scheme During Encryption Using Machine Learning

Reversible data hiding (RDH) is a recent research field of information security for secured digital data transmission. The advancements in communication technology and the invention of new medical robotics are very much useful in telemedicine applications. The transmission of medical image and electronic patient records (EPR) is a common process in telemedicine. The images captured by robots may need to be authenticated, the RDH schemes can be used to authenticate data and/or the owner of the data. In addition, the RDH techniques provide a way to embed EPR data into medical images before transmission. The EPR data extraction and recovery of the original image can be carried out by the receiver. This manuscript proposes a new RDH scheme to embed EPR data during image encryption process. A block-wise image encryption technique has been used in the proposed scheme to obtain the encrypted image with hidden EPR data bits. The novelty of the proposed scheme is that a support vector machine (SVM) based classification scheme has been used for data extraction and image recovery process from the encrypted image. Experimental study of the proposed scheme on standard medical images from OsriX dataset shows that the proposed scheme performs better than the existing schemes in terms of embedding rate and bit error rate.