Advanced WLDvG for medical image forgery detection
M. Arun Anoop, Palanisamy Karthikeyan, Ben Othman Soufiane, Shanmugam Poonkuntran · 2025
Medical images of breast cancer may be transmitted through internet; also, clues hiding that are embedded in those images can turn out to be a serious threat to the transmission of medical images. Physicians and medical professionals do not recognize such type of forgery and in some cases this can happen through digital image transmission. A few middle-aged or less experienced doctors may not recognize such minute changes in digital images, which might lead to wrong prediction. Little modification may affect image quality and the identification of image falsification that was carried out. Proving the authenticity of medical images is a task performed using stacked versions of machine learning (ML) and deep learning models. In a way, there is only a slight change in medical X-ray images that can seriously affect patient's health, and this may lead to wrong patient diagnosis and seriously threaten patient's life; it might also have a mental impact on few patients. Early detection of such type of forgery is necessary to help patients overcome such dangerous situations. Features are extracted based on variants of Weber local descriptors and TensorFlow hub's feature vector-based CNN pretrained models. Subsequently, the most intelligent algorithm like ELM with pipeline and linear model options are used to check the authenticity of images. The proposed algorithm is used for breast cancer digital images, accuracy detection is performed to show the effectiveness of methods, and finally the results are compared with other methods. This chapter mainly utilizes 17 supervised ML techniques along with Alexnet, RESNET, RESNET50, VGG CNN pretrained stacked models for the detection of medical image forgery. Elephant herding optimization and genetic algorithm optimization are the best methods used with high accuracy.