ADVANCING BIOIMAGING CORE SERVICES WITH ARTIFICIAL INTELLIGENCE
Paul Hernandez is a mathematician who develops algorithms for image analysis. Currently, he is pioneering the use of machine learning and deep learning techniques for image analysis. He is an imaging scientist awardee of the CZI cycle 2 call. His project aims to develop accurate and high-throughput AI algorithms to perform bioimage analysis and develop custom-made solutions for researchers. His efforts include developing and deploying open source software and training courses to reach the global community of imaging scientists.
What was the inspiration for your project? How did your idea for the CZI project arise?
Since I was an undergraduate student, I wanted to see how Mathematics could be applied to solve everyday life problems – something important to society. Since my MSc at Universidad Veracruzana I got involved in topics related to applied mathematics. In my PhD I developed my interest in Machine learning and Artificial Intelligence applied to bioimage analysis. After finishing my PhD I had the chance to join the Institute of Biotechnology (UNAM) where it became clearer to me how images are acquired and how they can be analysed. I realized it’s limiting when you want to manually analyse thousands of images, and the importance of automation became clear to me. In the last 8 years I’ve been working at the intersection of Biology, Physics, Mathematics and Computer Science. I think the project was already in the making in this sense, by the time the call came up.
At what point in this process did you first hear about CZI and how did you decide to apply for this call?
It was around 2020 when the CZI imaging scientist cycle 2 call opened. Dr. Chris Wood, the Director of the Mexican National Laboratory for Advanced Microscopy, is very well connected and deeply involved in this field. He mentioned this call to me and I realised that it was exactly targeted to what I had been working on for the previous 6 years at the institute. So the application went very smooth because the work we had done was very relevant.
Making things accessible when bridging disciplines is very important. What do you think about the role of this and community engagement for the advancement of microscopy and image analysis?
I think teamwork and clear communication startegies are crucial. We speak different languages – Mathematicians, Computer Scientists and Biologists- and we each are familiar with the discipline’s jargon. When we need to work as a team, this simple communcation can be difficult if we are not familiar with the jargon and there are no ‘translators’. We need to find a middle ground on how best to communicate and we have dedicated meetings to enable successful exchanges of information altogether. One of the things I’m trying to do in my project is develop code that is easy for Biologists to understand and that motivates further engagement.
How has the CZI support helped you reach your project goals?
It has been vital. The funding has given my work a big push, and a huge network of collaborators. The fact that I’ve been able to collaborate with so many people also was vital for my professional development. I think it allowed me to get now a position as principal investigator, where I can have students under my supervision. I’ve seen the need for image analysis methods, which is huge, in the field of biology and biomedical sciences, and we need more experts in this area. I’m not sure there are many scientists in Mexico working on AI-related projects applied to the biological sciences. So being able to supervise students means I can help form the next generation of computer scientists who are linked to the biomedical science field.
In general, what do you think is missing for microscopy and image analysis to be accessible to everyone in the world?
I think we have a lot of tools, but what we need is better dissemination and integration of the tools we have. There are many tools that are not being used to their full potential due to lack of awareness or lack of knowledge. For lots of the problems I help address, I often search on the internet whether software already exists that tackles those problems, and often it does. So we need dissemination events to talk about the tools that already exist, so these tools can reach the right people. I think that’s one thing CZI is achieving: they are bringing together labs from different disciplines and promoting the creation of workshops, courses and other training events to talk about the tools that exist and how to use them. I have given webinars at BINA, and other events such as I2K have been fantastic. I think the networks that CZI has helped fund such as BINA and LABI are a great way to connect people. We get monthly information on workshops, courses, research opportunities, funding, etc. The same is true of Mexican Bioimaging, which is allowing me to meet many people in Mexico who are working on microscopy.
Have you been able to collaborate more with the community since becoming a CZI grantee? And in addition to funding, what else do you feel is unique to the CZI program?
Absolutely. I have had the chance to collaborate with scientists like Federico Lecumberry and Leonel Malacrida. There are lots of connections we have already established and which we’re looking forward to continuing to foster. CZI has been vital for this. I think if I weren’t a grantee, I wouldn’t have met all these key people. Connections with other scientists like Thierry Pecot has helped me grow as a scientist, and it also avoids duplication of work, to know what has worked and hasn’t. So altogether community-building is huge for all sorts of reasons.
What is your biggest success during the CZI project?
I love being able to reach lots of people. Organizing workshops has enabled me to not just be dedicated to one single research group, but to expand and reach a lot more people with different levels of knowledge and interests. I love being able to organize the workshops on the work I do and the research I carry out. Having a workshop with 20 people for me is very satisfying. When I’ve taught virtual workshops with 50 people, it’s also incredible. Seeing people I’ve trained go on and succeed in generating more complex analyses and publishing their findings is something very satisfying to me. Moreover, it was always a passion for me to solve problems that are relevant to society and I think we are accomplishing this!
Where do you want to be with your project in 10 years? How do you feel AI will develop in the long term?
Ideally we will have something that no longer depends on us (mathematicians and computer scientists). I hope there will be software and other tools that are well established and easy for everyone to use – truly accessible. I hope all the tools we are developing now one day become as common as Office, Word, Excel or Power Point, but for the area of image analysis. I want to develop something like this, and something much more unified. I also see that young students today are more open to learn tools based on programming. People are now venturing into this area because they see the advantages to the work we do. This is a motivation in and of itself, to continue developing software that is much more accessible to biologists.
Check out our introductory post, with links to the other interviews here