The Imaging Working Group will support the application of data analysis and AI tools to understand disease mechanisms in dementia, and for translation to clinical use. There are a host of novel data science and machine learning methods for neuroimaging that have the potential to provide new insights and transform patient care.

We see particular opportunities in a multidisciplinary approach, combining neuroimaging data with other biomarkers. We will therefore work collaboratively across all the grand challenges and with other working groups.

The Imaging Working Group will begin by reviewing the current state of AI in neuroimaging for dementia research. Firstly, we will review methods to integrate data across imaging modalities, including brain, heart and body imaging. Secondly, we will review the development of clinically translatable AI applied to imaging data for transdiagnostic imaging markers across dementias, and for individual prognosis.

Supported by these reviews, we will provide standards for the use of multi-centre and multimodal imaging, image quality control and interpretation of imaging outputs from machine learning or data science methods. In addition, we will provide guidance on translating AI methods for clinical use, including the collection and sharing of appropriate datasets, validation methods, and regulatory issues.

To increase the multidisciplinary impact of neuroimaging in dementia we will:

  1. openly share pipelines for the application of machine learning to imaging data,
  2. establish standards for sharing neuroimaging data, in collaboration with the Dementia Platform UK,
  3. providing training and lectures on AI as applied to imaging.

We are excited that neuroimaging has a central part to play in data science applied to neurodegeneration, and that now is the time to harness the recent strides in neuroimaging analysis and AI for the benefit of patients.