AI-powered malware simulation of a medical imaging database

The dataset comprises medical imaging data that demonstrate the presence or absence of illnesses. used to simulate AI-based malware modulation, this database is paired with malware-modulated counterparts. By creating tampered images on the fly from the benign dataset using three mechanisms:Adversarial perturbations to input data that can cause data misclassification.Patch-level content edits by Copying-pasting or inpainting of small square regions (8–32 px) to simulate lesion insertion or removal.Metadata-consistent rescaling for random resize and crop variance. Each training batch is a duplicate of the original images.