Visiting Scholar
Andrija Stajduhar
During my stay in Montreal, I focused on advancing a neuron-centric analysis of the BigBrain dataset, bringing my previous experience in segmenting cells with data from the Zagreb Collection, and leveraging explainable AI to explore and validate cutting-edge methodologies. A key part of my work was a comprehensive review of the critical data availability for this research, which provided a clear direction for future analyses. I collaborated with a PhD candidate Alejandro Salinas Medina to update and refine the codebase, ensuring the computational workflows are robust and scalable. I obtained great support from Paule-Joanne Toussaint, who instigated this collaboration, facilitated key introductions, provided strategic guidance and scientific inputs, as well as from Claude Lepage, who contributed technical solutions essential for advancing the computational aspects of the project. Together, we began developing a pipeline to validate cellular segmentation and quantification methods from our collaborators in Jülich against human-labeled data. This effort is a crucial step toward ensuring methodological accuracy.
In terms of progress, I contributed to successfully create a cellular density map for a 6x6x6mm cortical block, laying the groundwork for further feature extraction. Testing state-of-the-art segmentation methods has revealed areas requiring refinement, and we are moving forward with plans to validate these methods. The insights gained here directly align with the broader goals of the project, which include automating large-scale histological data analysis and enhancing our understanding of cortical organization.
I attended the Montreal AI and Neuroscience (MAIN) conference, where I established valuable connections with researchers and clinicians. These interactions have opened new doors for collaborative work and provided fresh perspectives on the clinical applications of our research. I also initiated collaborations with researchers at MILA and The Neuro, who bring complementary expertise to this project, but also create other research initiatives of my interest. Additionally, I gained insights into how medical education is organized at McGill university, which could inform our teaching and training programs at my home university.
A highlight of my stay was presenting our recent work at the Feindel Brain and Mind Seminar series with a talk titled Advancing Neuron-Centric Analysis of Human Cortical Cytoarchitecture with the BigBrain. This was an opportunity to disseminate our results, share our methodological advancements, and engage with experts in the field.
Looking forward, we have established a strong framework for sustained collaboration. This includes the validation of segmentation methods, creation of new datasets and refinement of existing ones, and co-authoring publications to share our progress with the scientific community. The support and resources provided by The Neuro have been invaluable, and I’m eager to continue building on this years long partnership also in the years ahead.
