Hello Marek, I’m glad to meet you! Where are you from?

I am from Poland – currently working both in Poland (AGH University of Kraków) and Switzerland (HES-SO Valais).

What is your background and your professional activity?

I have a PhD in Biomedical Engineering & Computer Science from AGH University of Kraków and I am currently working as PostDoc in HES-SO Valais. I have worked in the field of medical image analysis, computer vision, machine & deep learning for more than 7 years so far.

Why did you decide to participate in the France-BioImaging challenge “Light My Cells”?

Taking part in scientific challenges is a great way to explore previously unknown subfields and gain valuable experience in new topics. When I came across the challenge on the IEEE ISBI website, I immediately recognized it as an opportunity to deepen my expertise in biological imaging and expand my research network.

What was the most challenging part of the competition for you?

As in every challenge – to fully understand the goal, the data, and the associated challenges. The most time-consuming part in scientific challenges work is to explore the dataset and understand which challenges related to the data are the most important, e.g. in the Light My Cells challenge it was connected with significant dataset imbalance, both at the study and organelles level.

How did you manage your time during the competition?

I started with exploring the data and thinking how the problem should be addressed. The process took me more than half of the time I spent on the challenge. Then, I implemented and debugged the training and evaluation scripts. Finally, I queued all ablation studies on our supercomputing platform (ACK Cyfronet Athena) and chose the best model.

Do you have any projects or aspirations related to imaging or research after this competition?

The competition improved my experience related to image-to-image translation tasks that could be further used in different downstream applications I am working on, e.g. MR-to-CT translation in radiology. I continue to work both in digital pathology and radiology where I develop novel deep learning techniques.

What are your thoughts about Challenge 2 “Fuse My Cells”?

Definitely the most difficult aspect of the Fuse My Cells challenge is the timeline – there is only 1.5 months to develop and evaluate the solution.

Do you have any advice for the participants of Challenge 2?

I suggest starting with exploring the data and understanding the associated challenges. Then, it is wise to explore the current state-of-the-art to understand the best-performing solutions and improve them. The practical part related to developing the scripts and  performing ablation studies is usually significantly less influential.

Thank you very much for your time, Marek! I’m sure your testimony will be useful for the participants of Challenge 2.

For those interested in taking part in the “Fuse My Cells” challenge, find more information here!