Detection of Copy Move Forgery in Medical Images Using Deep Learning

Since the advancements in technology and IT has revolutionized the world, digital images have come out with crucial importance. With the fruitful advancements and purposes, the authenticity and security breaches in digital images are simultaneously increasing because many editing software and tools give easy access to manipulate and temper the images, resulting in the change of complete information. Copy Move Forgery is the simplest way of tempering images in which an object is copied, removed, and replaced in the same image. As the medical field is too sensitive and even a minor manipulation can produce disastrous results, this study proposes an algorithm specifically designed to detect copy move forgery in medical images, especially when the world has gone towards telemedicine due to the outbreak of COVID-19. The proposed algorithm is based on CNN working on the whole image. The algorithm works in three phases, i.e., pre-processing, feature extraction, and classification. The proposed algorithm has given the accuracy of 89 percent on the dataset that has been created due to the publicly non-availability of forged medical images dataset. The dataset includes the images from abdominal, lungs, transverse view of lungs, chest abdominal, lungs transverse, lungs ap, vertebrae, and transverse heart.