Artificial intelligence (AI) and machine learning (ML) developments applied to neuroimaging in dementia are rapidly advancing. This paper provides a roadmap for clinically meaningful translation of this to healthcare systems. Firstly we discuss the current state of neuroimaging in dementia. Secondly we discuss the potential benefits of AI/ML applied to neuroimaging. Thirdly we discuss the pressing need to translate these advances in a clinically meaningful way for improved diagnosis, prognosis and stratification. Fourthly we discuss the translational barriers, regulatory requirements, and finally the potential solutions and roadmap for the development of clinical tools using AI/ML technologies in brain imaging to memory clinics.