Meeting with: Trang Le, one of the challenge “Light My Cells” winners!

On the occasion of the launch of France-BioImaging’s new challenge, Fuse My Cells, we reached out to the winners of the previous edition (Challenge – Light My Cells). Today, we invite you to meet Trang Le, PhD student in Bioengineering at Stanford University.
Hello Trang, I’m glad to meet you! Where are you from?
I’m originally from Hanoi, Vietnam. Then I moved to Europe for my bachelor and my master study, and now I’m finishing my PhD in the US at Stanford University.
What is your background and your professional activity?
I’m a PhD student in bioengineering focusing on machine learning for image analysis and modelling of single cell spatial proteomics, towards an AI virtual cell model. I’m also interested in citizen science and accessibility of AI models in biological research.
Why did you decide to participate in the France-BioImaging challenge “Light My Cells”?
I heard about the challenge from my advisor after she returned from a conference. Originally, I wanted to test out a different strategy of modelling complex joint distribution, which did not work for the task as intended. In the end, I opted for a more simplified solution, keeping only the essential network components and luckily that was enough.
What was the most challenging part of the competition for you?
Surprisingly it’s managing the submission format!
What are your thoughts about Challenge 2 “Fuse My Cells”?
It seems like an interesting and useful task, especially with the growing amount of multiview microscopy data. Ultimately, the developed models should learn meaningful biological features and generalize well to new imaging conditions. The challenge’s design and evaluation metrics will be crucial in guiding progress toward practical and robust solutions.
Do you have any advice for the participants of Challenge 2?
A few things come to mind:
- Spend some time on dummy submission earlier
- Think about the generalizability of your model (and create your validation set fairly)
- This competition is quite pressed for time, so keep your preprocessing simple at first
- And have fun!
Thank you very much for your time, Trang! I’m sure your testimony will be useful for the participants of Challenge 2.
As the first-place winner of the competition, Trang had the opportunity to present her solution, VQGAN, to the entire France-BioImaging community at our last Annual Meeting in Strasbourg. We are delighted to share this moment with you!
For those interested in taking part in the “Fuse My Cells” challenge, find more information here!