Use of artificial intelligence in multiple sclerosis imaging

The goal of this chapter is to provide an overview of artificial intelligence applications in the context of neuroimaging individuals with multiple sclerosis. First, we will introduce the most widely used artificial intelligence algorithms as well as the methods used to assess the performance of artificial intelligence algorithms. Second, we will provide an extensive review of the potential benefits of artificial intelligence applications in multiple sclerosis neuroimaging including improving MRI protocols and image quality, tissue and lesion segmentation, classification of multiple sclerosis lesions into different subcategories, as well as diagnosis, prognosis, and clustering of multiple sclerosis patients using multimodal neuroimaging techniques. In addition to the potential benefits of using artificial intelligence in multiple sclerosis neuroimaging, we will discuss ethical considerations and potential pitfalls of artificial intelligence, including model generalization, perpetuation biases, and issues with explainability. We end with recommendations and future directions for the successful application of artificial intelligence in clinical settings.