A Blend of Deep Learning and Machine Learning based Tamper detection of medical images followed by localization and self-recovery

Telemedicine provides a means of remote clinical services, via real-time duplex communication between two medical practitioners, using electronic audio and visual means. With the technological advancements in the medical field and image processing, there is an exponential rise in the medical data taken by doctors to diagnose the disease. The medical data comprises medical reports and medical images which are transferred through covert channels that may be subjected by tampering. Tamper detection and localization are highly required to safeguard the integrity. As the images are exponential, it became very laborious for the doctors to analyze each and every image. Machine learning approaches based on multi-class SVM, Random Forest, and XGBoost are used to train attacks of various tampered photos, relieving the doctor of the need for manual detection. Based on the detection of a tampered image, localization and recovery are carried out to recover the original image.