STEM4Brit: Sharing cutting-edge healthcare engineering research with British policy makers
Earlier this month Benjamin Davidson participated at STEM4Brit, an event to promote the sciences to UK Parliament.
Scientific advancement and innovation are crucial elements of the UK’s Industrial Strategy. One event that helps bringing together scientists and policy makers is STEM4Brit, a poster competition for early career researchers hosted in Westminster since 1997. Aiming to increase the support and promotion of the STEM field, the contest is highly acclaimed within the community; this year only 35% of the applications were chosen to present at Parliament. Benjamin Davidson, PhD candidate in our UCL CDT for Medical Imaging programme was selected to be part of STEM4Brit 2018 event and shared his thoughts on the experience.
What research project did you present at the STEM4Brit poster competition?
My research focuses on developing useful clinical tools, that can be used to automate medical image analysis. Specifically, this work includes the creation of an automated workflow for the analysis of Adaptive Optics Scanning Light Ophthalmoscope (AOSLO) images. AOSLO imaging is used to provide a non-invasive, microscopic view, into the living eye. The images the AOSLO produces contain extremely valuable information that allows us to better understand disease, as well as investigate the therapeutic effect of experimental gene therapies, which in the future we hope can cure currently untreatable, blinding conditions. The downside of using this tool, however, is that image analysis is laborious, and therefore expensive and slow. By automating the processing larger, cheaper clinical studies can be carried out. More clinical trials for less money means faster development of treatments.
The project has already had major successes, culminating in saving the clinicians up to 4 hours in analysing single patients’ datasets. This was achieved by applying classical computer vision algorithms, such as feature based image stitching, and more recent techniques from deep learning, such as multidimensional recurrent neural networks for image segmentation. Each of these techniques is currently deployed in Moorfields Eye Hospital, and has been incorporated into the clinical workflow.
What was the most unexpected learning point from this experience?
The research project I am working on is highly interdisciplinary, encompassing advanced mathematics, engineering, and ophthalmological knowledge. It was very satisfying and challenging to engage with each of these groups and hear about their distinct approaches to research associated problems and solutions, as well as which were the different aspects of my poster that they found interesting.
Have you engaged with any policy makers at this event and what do you think about policy making and science?
At STEM4Brit, I had the pleasure of discussing with Ms Cathering West, who is my M.P. for Hornsey and Wood Green. In what concerns science and the public, I think there needs to be a greater collaboration between the two. With the recent rise in distrust in the field, one common thought seems to be that it is because the public doesn’t understand science. Instead, I feel it is researchers’ obligation to engage with the public in an interesting, accessible and apolitical manner. Improving public opinion of science will hopefully attract more focus on STEM, which can ultimately lead to more evidence based approach to policy making.
What do you think were your success factors in having a successful submission?
My area of research is highly applicable to the UK economy, and has produced some strong results, despite being in its infancy. Automatic medical image analysis is a growing area, and has the potential to dramatically increase the efficiency of the NHS. My project could be considered a case study of how automation could look, and will uncover important learning experiences, which are important to share if we want to develop useful tools for doctors to use.
How closely do you collaborate with clinicians and industry and what are the next steps for your research?
I work closely with my clinical collaborators, as by partnering with them I am better able to satisfy specific needs which are sometimes less interesting from an academic point of view, but could save hours of work in the hospital. For example, the current bottleneck in AOSLO processing is that clinicians have to fill out spreadsheets by hand. This can be automated in 10 minutes, and saves hours of work, but without close collaboration this sort of everyday problem would never be addressed.
My main aim as part of the UCL CDT in Medical Imaging programme is to make computational tools that have real clinical impact. Any cutting-edge AI, deep-learning, computer vision 2.0 method we may develop during this process comes to me as a bonus by-product.
Photo Credit: STEM4Brit Twitter