The Crick BioImage Analysis Symposium (2024): conference overview

Have you missed this year’s Crick BioImage Analysis Symposium (CBIAS2024), or you want to know how this conference went to apply for next year’s edition? Here is our recap of this conference featuring impressions from 3 microscopists coming from different research backgrounds:  

Zeinab Rekad, postdoctoral Research Associate, biologist, super-user and here for the 3^rd time 😊

Participating in the Crick Bioimage Analysis Symposium for the third time has confirmed my prior impressions that this conference is one of the most engaging, and enriching event the field has built thus far. And as the previous years, a variety of ‘hot topics’ in the image analysis field were discussed, including:  – AI in bio-image analysis, from smart microscopes to challenging multimodal datasets analysis. – But also, the importance of data management, data sharing and storage. – As well as a collective update on the perspectives and experiences around what is it like to be an image analyst. 

AI in Bio-Image Analysis: Opportunities and Challenges 

One of the most stimulating aspects of the conference, in my opinion, was the depth of discussions around the rise of AI usage in image analysis. Hearing directly from experts as Assaf Zaritsky, Rolf Harkes, Laura Wiggins and many others who are developing AI-based pipelines offered a refreshing perspective. Indeed, as we celebrated the rapid advances and possibilities AI brings to bio-image analysis, the discussions were balanced. The panel talk led by (Amy Strange, Assaf Zaritsky and Lydia France) was particularly striking, addressing issues such as over-reliance on AI, biases in data, and the challenges of reproducibility. These conversations added much-needed moderation to the often-overhyped narrative surrounding AI. That said, the panel and audience also shared optimism about the future of the field and there was a consensus on the importance of equipping the next generation of bioimage analysts with strong foundations in both traditional and AI-assisted analysis as well as the need for establishing best practices and cultivating a critical yet forward-looking mindset for the successful long-term implementation of AI in image analysis.  

Big Data…and an even bigger problem of handling / storage 

Another focal point (pun intended) of this year’s conference was the growing challenge of handling hyper-volumetric datasets, which now reach petabyte scales. Discussions led by Aastha Mathur, Sophia C. Maedler and others touched on the technical hurdles of storage, sharing, and collaborative analysis of such massive datasets. Beyond storage, the importance of accurate annotation and rigorous quality control was underscored. This theme resonated strongly with me as it exemplified the dual nature of technological progress: while we can now generate and analyse unprecedented volumes of data, the infrastructure and workflows required to handle them are struggling to keep pace. These conversations emphasized the importance of a collaborative effort across disciplines to address these pressing challenges. And I would not be surprised to see these conversations carried out over the next issue of the conference along with the emerging realisation that image analysis will soon cross path with other ‘big data’ analysis domains such as OMICs.  

Bio-Image Analysis: the humans behind microscopes and computers 

As in previous years, the symposium also addressed the “human side” of bio-image analysis. Topics such as the recognition and appreciation of bioimage analysts within institutions, funding bodies, and the broader scientific community were discussed at length. It was encouraging to see these conversations persist, as they highlight the critical need to ensure the sustainability of this field and the well-being of its contributors. What struck me most during the Image Analysis Community Initiatives panel (led by Tom Slater, Aastha Mathur, Anna Klemm, Stephen Cross) was the sense of community. From junior researchers to senior experts, everyone participated in an open and inclusive manner. To me, this level of interaction, which enables newcomers to engage freely with established professionals, is a hallmark of this symposium’s success. It was also heartening to see a seminar theatre filled with passionate and (bit geeky 😊) researchers till the very end of the conference.  

Final Thoughts 

The field’s potential is immense and talks from Christophe Leterrier, Guillaume Jacquemet, Gebhard Stopper and many others, truly reflected how image analysis is now evolving from “turning images into numbers” to generating actionable data that unlocks previously inaccessible knowledge, transforming our approach to biological and medical research.  

To the organizers, sponsors, and participants: thank you for your dedication, hospitality, and curiosity. Whether you are new to the field or a seasoned expert, I cannot recommend this conference highly enough. The opportunities for networking, learning, and contributing to field development are unmatched. 

For me, this year’s symposium was a reminder that as we advance technologically, the strength of our community and our commitment to thoughtful, balanced progress will be what drives the field forward. I look forward to participating again next year and as long as this symposium will exist 😊

David Gaboriau, Imaging Facility manager, cell biologist and Imaging Scientist, here for the 5th time 😊 

At the end of November, the Crick BioImage Analysis Symposium, or CBIAS, was held at the Francis Crick Institute. I’ve attended all five editions, and the latest one did not disappoint: informal and friendly, and with an excellent line-up of leaders in the field and early career researchers, CBIAS is a personal high point in the year and I always very much look forward to it. 

Over two days, this year’s main themes were smart microscopy, what AI can and can not do, applications of bioimage analysis in biomedical research and bioimage analysis community initiatives, supplemented by lively panel discussions about generative AI, how to build communities and what is means to do bioimage analysis. 

This year, the poster session was split in two, giving ample time for browsing and discussion. 

In the Smart Microscopy session, we heard about strategies to acquire the highest quality data when it is needed: Suliana Manley described how using an imaging strategy with an event-driven acquisition controller enables imaging of rare events in mitochondria, in this case fission. As these are preceded by a recognisable precursor, the microscope knows when to switch to a much faster acquisition mode to capture the fission and return to slow mode to minimise photodamage. 

Transmitted light imaging such as phase contrast can also be used, as shown by Khalid A. Ibrahim, who is using AI-driven imaging to detect and assess protein aggregation, here taking advantage of a label-free technique. 

As a community, we are keen on using metrics to evaluate the quality of fluorescent images, but we were reminded by Richard Marsh about how commonly used quality metrics perform badly as they are very sensitive to non-structural parts of the image (i.e. noise). 

After lunch and a first look at the posters, the session on what AI can and cannot do started with Laura Wiggins showing how AI can help classify and understand their topology of the complex shapes taken by DNA loops imaged by Atomic Force Microscopy. 

AI models are often compared to a black box, as there is no way to understand how the model made a particular decision. But, by using image-to-image transformation models, Assaf Zaritsky is able to dig inside the box and start explaining how a model created a certain output. 

Then followed a very interesting panel discussion on generative AI. The consensus is that it is crucial to improve the general public’s understanding of what Large Language Models are and how they work. Similarly, students need to be taught now about AI models, whether they use them for writing essays or segmenting cells. Importantly for bioimage analysis, students should also be shown classical approaches and the theory underpinning them, so they can make informed decisions on which approach to use.  

It was now time for the second poster session, followed by a social event nearby. 

Day 2 started with Christophe Leterrier diving into the ultrastructure of the axon and using wonderful multimodal correlative imaging to identify novel actin structures that surround the clathrin-coated pits active in endocytosis. 

Gebhard Stopper showcased an automated pipeline to allow tracking and monitoring the killing of cancer cells by killer cells. 

Guillaume Jacquemet presented analysis tools developed in his lab and how they can be used in a very neat experimental system where tumour cells flowing over a layer of endothelial cells are imaged with brightfield to study extravasation.  

The lovely thing about this community is that most tools presented are open-source and sharing is a big thing, so you normally come back from CBIAS with quite a list of new tools to test and use for your own projects and many new ideas. It’s really invigorating! 

The afternoon was dedicated to the notion of community in microscopy and bioimage analysis, presenting existing networks and exploring what a community is and how we can establish and grow them to include and support students and researchers.  

The CBIAS has reliably been my favourite conference. It hosts a very friendly, approachable and supportive bunch of people, reminiscent of the NEUBIAS events, where you can meet old and new colleagues, find out about the newest innovations in the field, all in a relaxed and welcoming setting. I really recommend it to anyone interested in bioimage analysis, beginner or expert. A big thanks for the organisers who are reliably putting on one of the most enjoyable and high quality conferences around! Can’t wait for the next one!

Fiona Love, Impressions from a CBIAS newbie: postdoctoral cell biologist, here for the first time! 

First Thoughts 

I really wasn’t sure what to expect from CBIAS.  As a neuroscientist and cell biologist I’ve been doing bits of bioimage analysis in some form or other for a while now, but I’ve recently started working with some more complex data, and I wanted to see what kind of tools and technologies others in the field were using.  I was a bit afraid everything would go way over my head — I’m pretty handy with code and I’ve written a few ImageJ macros for simple quantification tasks, but I wasn’t sure how far that would get me with the ‘real experts’. 

 As it turned out, I definitely didn’t need to worry!  With a mix of biologists and computer scientists in the audience, all the speakers did a fantastic job making their talks accessible to any background.  I feel like I got a good overview of the current state-of-the-art in bioimage analysis, and I learned a lot that I will be taking back to my own work! 

New Discoveries 

After a quick sprint through St Pancras Monday morning (my train from Cambridge was cancelled), I managed to arrive just in time for the first session on smart microscopy and was immediately blown away.  Suliana Manley started off the session with an excellent talk that highlighted how smart microscopy worked and the kind of biological insights it could give us.  I’m familiar with super-resolution microscopy and other ‘advanced’ imaging techniques, but I hadn’t realised there was so much work going into getting more out of ‘conventional’ microscopy.  I found this really impressive, and I’ll be keeping it in mind for my future experiments. 

 I also hadn’t come across holotomography before, but the data from that looks amazing!  It’s a completely different approach to the multispectral fluorescence imaging I’m doing, but one that could be a great complement once I get a bit more data.  Definitely something I’ll be keeping an eye on! 

Discussion Highlights 

I found the first panel discussion on (generative) AI immensely useful.  I’m a bit of a sceptic on this front — I still don’t use ChatGPT — and it was nice to hear that my reservations aren’t completely unfounded.  More importantly though, there was some great nuanced discussion on how we can decide where AI use is justified, how we should validate its output, and how to communicate the use of AI in science with members of the public. 

 I also had some great personal discussions during the breaks and poster sessions.  It was nice having such a range of people to talk to across so many different areas of biology and computer science, as so many of our analysis problems ultimately have similar solutions.  I came to CBIAS to get an idea of what tools people were using for their image analysis, and I got advice on different software platforms, tips for organelle segmentation, and links to useful analysis packages.  Mission accomplished! 

A New Community 

What really struck me the most about CBIAS is how friendly and collaborative the bioimage analysis community is. I had a great time over these two days at the conference, but I’ll also be keeping up with the contacts I made and the wider community online through the rest of the year.  And I’m already looking forward to CBIAS next year!

Final note: we all wait for you at the next edition of CBIAS, but for now if any of the talks mentioned above have caught your attention, feel free to check the symposium’s YouTube page where talks will be uploaded beginning of 2025.

(2 votes, average: 1.00 out of 1)