Computer Vision News - April 2021

talking about fetal screening, absolutely not. It’s an incredibly difficult question. It’s a balance. There’s a huge amount of good that could be done, but as with so many things in science, you have to make sure you use it for the good. I’ve known you for only a few minutes, Emma, and you told me several times that the goal of your work is to understand things. When I asked you about their practical application, like fetal surgery, it sounded like that is not exactly what you are aiming at. Can we go beyond understanding and find some practical applications for your study? You’re talking about the difference between work that has an immediate path to the clinic and work that has a long-term path. In terms of immediate impact, we’re looking at tools for detecting signs of epilepsy for surgical planning for epilepsy. Are we anywhere near being in theposition todetect really early signs of dementia from MRI? No. When you start getting into complex psychiatric heterogeneous neurological disorders, then I think the question becomes much more complicated. Do I have a practical goal in mind? Yes. Am I tackling more difficult problems than can be immediately delivered to the clinic? Yes. You talk about detection as a practical application, but in a non-scientific mind like mine, the only practical application is something that you intervene on. Detecting doesn’t seem like a practical application, to someone like me. Let’s go back to the basics, for example. Let’s say that through our analysis, we are able to detect, at an early age, which babies are more likely to be at risk of long term cognitive impact. Then you have a set of biomarkers from which you can use in clinical trials of potential treatment. The same for neurodegenerative diseases. The idea is that you want to be able to detect early signs of disease that can subsequently be used as biomarkers for clinical trials and interventions. Also, to develop causal models of disease. The direction that we’re going in now is to build a model of how a brain might change over time and create, what they call, counterfactuals, from which one might be able to pose different questions, and through that, understand different disorders. You are dedicating the best years of your career to this subject. What would be your dream achievement? What we want to be able to do is to detect an individual with a high level of 34 Emma Robinson “We want to propagate that information using a combination of novel, portable technologies, apps, wearables, and all sorts of things…”