Our mission is to create methods to fit the science and not make science fit the methods.
Specifically, we are interested in the following questions that have clinical relevance:
1. Neurodegeneration—How can we develop software frameworks that can assist dementia screening in various real-world settings?
2. Digital pathology—How can we build clinical-grade software tools to complement the clinical workflow?
We are also interested in the following frameworks that have computational relevance:
1. Domain generalization—Development of deep neural networks that can generalize well across multiple data cohorts.
2. Representation learning—Construction of efficient neural models on high resolution data to process local and contextual information.