Computer Vision News - April 2021

Did you choose the brain because of the Human Connectome Project or did you decide it independently and you joined the project after that? I was working on the brain beforehand, so the whole time I’ve worked on the brain. Originally, I joined a lab at Imperial College modeling baby brains. I did my PhD between there and the computer department at Imperial. I’ve kind of come full circle. I’ve always worked with brains. I don’t have any interest really in working with other organs. Everyone tells me to diversify. I think it’s a big enough problem in itself. It’s really important. You need a lot of specialist knowledge. You cannot treat brain data in a data-agnostic way. You need to have some understanding of the constraints of image acquisition and neuroscience to build informed models. You have to have sensible constraints. You’ve got very high dimensional data space, very complex multimodality data. Knowing something about the underlying neuroscience can really help. Is the reason for learning about the fetal brain to be able to one day intervene before the birth of the child? In my opinion, this is not about trying to detect whether someone might have some condition. We’re never going to be able to be that certain. Also, this is not a purely genetic thing. Just because someone has a set of genes, the basic initial blueprint that may indicate that there might be a possible disability that would develop, that’s never going to be set in stone. A lot of these complex conditions have significant environmental influence. My interest is trying to understand, from cognitive neuroscience, the brain and maybe help identify early signs to help more people at the earliest possible stage. If you’re 242 Women in Computer Vision “You must absolutely understand neuroscience!” METRICS Lab at King's College London