Tell us a little about yourself, and what attracted you to working at the University of Oxford?
I loved sciency things since as a kid, building science and electronic projects in my bedroom, setting light to my bin, getting numerous electric shocks from the mains and from a huge capacitor from a TV I was taking apart. I went to a state school near where I grew up in Woking, outside of London. Although they had never had a student accepted at Oxbridge before, some of the teachers were fantastic at encouraging and supporting me. They gave me extra help with preparing for the Oxford entrance exam and I was accepted to study Physics at Keble in 1986. I immediately loved everything about Oxford - both the town and the university. I quickly realised that I wanted to have a career doing research, ideally mixing with lots of interesting and clever people, and in a beautiful place.
How did you get to where you are today? Can you tell us more about your career path?
So I finished my undergrad, and knew I wanted to stay in science, but wasn’t sure in what direction. While I was still dithering, I suddenly found this was solved for me when I was offered a funded DPhil spot with Mike Brady in his Robots group. My next three years were spent in a dark underground basement developing image processing for autonomous vehicles - a reality now but a very futuristic prospect back in those days. Loved every minute! For reasons I can’t explain, I am a little proud of the fact that my very first journal paper still has zero citations. After that, I was a bit lost regarding how to find a postdoc in academia, and ended up continuing the same work in industry.
In 1996 I was asked to setup an image analysis group as one of four founding academics at the Oxford Centre for FMRI of the Brain - FMRIB - which was our name for the first 15 years before we became WIN and then OxCIN. It was a no-brainer - in part as I so badly wanted to move back to Oxford. 30 years later, I’m still in the same job, whereas another FMRIB founder, Irene Tracey, is the VC! But really, I love this job - doing research and working with amazing people. At the start I was utterly clueless about how to build a group and supervise students, but we got very lucky. Some of our first students and post-docs turned out to be incredible - people like Mark Woolrich, Christian Beckmann, Tim Behrens and Mark Jenkinson. They all became world leaders in their own parts of the field and we continue to work together. Then as the group gained traction, new leaders like Saad Jbabdi, Jesper Andersson and Tom Nichols joined, to create a really remarkable team spanning huge breadth in analysis. And while they're amazing in terms of the science, the best part is that we're all good friends - a bit like family.
Can you give us a brief overview of your research?
FMRIB Analysis has developed exciting new approaches for understanding the brain, including modelling white matter brain connections probabilistically, and learning how to warp the cortex from subjects to another so that their areas of brain function are well aligned.
Ten years after starting the Analysis group, we were invited to lead major parts of the NIH flagship Human Connectome Project, which aimed to provide the most comprehensive mapping to date of human mesoscopic brain connections. We also lead the brain imaging in UK Biobank, which is by far the largest imaging study ever, with 100,000 participants. Both of these projects have provided huge insight into unmet needs in neuroimaging and been a major force guiding how we think about new analysis approaches and tools.
One research area that I have had a close personal involvement in over the past decade or so is “resting-state” fMRI, where the subject rests in the scanner. So this data represents spontaneous “background” activity, from which we can estimate a rich model of their brain connectivity and dynamics. One of our first major pieces of work in this space was showing that you really can measure the same networks in this resting state to those you identify by giving subjects explicit tasks. In more recent years, I’ve also gotten very interested in using population-level imaging data from UK Biobank to investigate how different parts of the brain age in quite different ways, and in finding genetic influences on the brain. During the pandemic, Gwen Douaud and I took advantage of a unique opportunity provided by the UK Biobank resource: the ability to directly observe pre-pandemic vs post-COVID-19 effects in the brain and demonstrate damage even for people experiencing mild symptoms.
You have put a lot of effort into neuroimaging software - can you tell us about that?
It was clear from the start that our analysis research would lead to new software tools. What was less obvious was whether we should just use those tools in-house or make a big effort to create a comprehensive and well-supported software toolbox that we disseminate openly to the wider community - what became the FMRIB Software Library (FSL). It may seem odd to early-career researchers now that we had to think really hard about whether there was even appetite for open-source tools from the broader community, but open science was in very early days then. We knew it would be a lot of work and that would take time away from the next exciting research project.
But it has really paid off, with respect to the impact we have been able to make on the field. FSL quickly became very widely used worldwide; for example, the core “methods” papers total well over 100,000 citations. Initially, FSL tools were generated by the group’s researchers, but over time we were fortunate enough to add a dedicated team for the software side, and pro coders like Matthew Webster and Paul McCarthy have done an amazing job expanding the software, and keeping it (and the researchers!) on the straight-and-narrow path of good coding practice.
You won the 2025 Glass Brain Award, given once a year by the Organization for Human Brain Mapping to recognize lifetime achievements by a leading neuroimaging researcher. How do you feel about winning The Glass Brain?
The Organization for Human Brain Mapping has always been our main conference, so this is a big honour. And, we just realised that 2025 marks the 25th anniversary of the very first FSL software release, so that’s nice timing! I know it can sound like a cliche, but in our case it is absolutely true that the award reflects many years of hard, thoughtful and creative work by the whole group, not just one individual. I can’t even take credit for being the Analysis group leader, because we have had many people leading analysis groups and projects over the years, quite independent of my input. I’ve been very lucky to be part of this.
What’s next for you and your research?
One big ongoing project for me personally has been the creation of a new journal, Imaging Neuroscience. The leading journal in our field had been NeuroImage, published by Elsevier. We managed to convince them to make the journal completely open access. But there was then huge concern about the massive profits being made by commercial publishers from OA journals because of the Article Processing Charge (APC). I became Editor-in-Chief for NeuroImage in 2022, and we tried to engage with Elsevier, hoping to reduce the APC from $3450 to under $2000. They were unwilling, and so all 43 editors resigned en masse and launched Imaging Neuroscience, an open-access non-profit journal with MIT Press (APC $1400). The announcement was met with wonderful enthusiasm and support from the scientific community (2 million views of the announcement within one week). In just over 2 years, Imaging Neuroscience has already published 750 papers and is widely considered by the community to now be the field’s top journal. Leading this has been hugely rewarding!
In FMRIB Analysis, a major new research goal is to integrate our analysis tools together more tightly with each other, while making them much easier to use by neuroscientists. Not exactly like a black box, because we still need imagers to engage deeply with their data (everyone who has been on our annual FSL Course will remember the looming Eye of Sauron demanding “look at your data!”). But as neuroimaging has matured, it makes sense to think about how we can make it more accessible to people who may want to use imaging but without becoming a deep expert: to make the main analysis easier, while also giving researchers the ability to ask interesting and novel questions of the data. It’s a bit of a nerdy goal, but it could significantly change the way imaging is done. For example, this would be crucial to providing an “FSL Clinical” that could be deployed in hospital settings. A lot of this will be based around our re-invention of the analysis pipeline that Fidel Alfaro-Almagro created for UK Biobank, which everyone in the group is contributing to.
And of course we will continue to push the boundaries of our brain modelling. For example, it is pretty shocking that there is still not yet a single comprehensive high-resolution “model” of the full dynamic brain connectivity data that we can get from imaging! Several people are working hard on new approaches for this (including of course people at OHBA led by Mark Woolrich), and I hope we’ll see some big progress on that in the next few years.