Simulating images representative of neurodegenerative diseases is important\nfor predicting patient outcomes and for validation of computational models of\ndisease progression. This capability is valuable for secondary prevention\nclinical trials where outcomes and screening criteria involve neuroimaging.\nTraditional computational methods are limited by imposing a parametric model\nfor atrophy and are extremely resource-demanding. Recent advances in deep\nlearning have yielded data-driven models for longitudinal studies (e.g., face\nageing) that are capable of generating synthetic images in real-time. Similar\nsolutions can be used to model trajectories of atrophy in the brain, although\nnew challenges need to be addressed to ensure accurate disease progression\nmodelling. Here we propose Degenerative Adversarial NeuroImage Net (DaniNet)\n--- a new deep learning approach that learns to emulate the effect of\nneurodegeneration on MRI by simulating atrophy as a function of ages, and\ndisease progression. DaniNet uses an underlying set of Support Vector\nRegressors (SVRs) trained to capture the patterns of regional intensity changes\nthat accompany disease progression. DaniNet produces whole output images,\nconsisting of 2D-MRI slices that are constrained to match regional predictions\nfrom the SVRs. DaniNet is also able to maintain the unique brain morphology of\nindividuals. Adversarial training ensures realistic brain images and smooth\ntemporal progression. We train our model using 9652 T1-weighted (longitudinal)\nMRI extracted from the Alzheimer's Disease Neuroimaging Initiative (ADNI)\ndataset. We perform quantitative and qualitative evaluations on a separate test\nset of 1283 images (also from ADNI) demonstrating the ability of DaniNet to\nproduce accurate and convincing synthetic images that emulate disease\nprogression.\n