Brain imaging has allowed neuroscientists to analyze brain morphology in genetic and neurodevelopmental disorders, such as Down syndrome, pinpointing regions of interest to unravel the neuroanatomical underpinnings of cognitive impairment and memory deficits. However, the connections between brain anatomy, cognitive performance and comorbidities like Alzheimer's disease are still poorly understood in the Down syndrome population. The latest advances in artificial intelligence constitute an opportunity for developing automatic tools to analyze large volumes of brain magnetic resonance imaging scans, overcoming the bottleneck of manual analysis. In this study, we propose the use of generative models for detecting brain alterations in people with Down syndrome affected by various degrees of neurodegeneration caused by Alzheimer's disease. To that end, we evaluate state-of-the-art brain anomaly detection models based on Variational Autoencoders and Diffusion Models, leveraging a proprietary dataset of brain magnetic resonance imaging scans. Following a comprehensive evaluation process, our study includes several key analyses. First, we conducted a qualitative evaluation by expert neuroradiologists. Second, we performed both quantitative and qualitative reconstruction fidelity studies for the generative models. Third, we carried out an ablation study to examine how the incorporation of histogram post-processing can enhance model performance. Finally, we executed a quantitative volumetric analysis of subcortical structures. Our findings indicate that some models effectively detect the primary alterations characterizing Down syndrome's brain anatomy, including a smaller cerebellum, enlarged ventricles, and cerebral cortex reduction, as well as the parietal lobe alterations caused by Alzheimer's disease.