Interview with Eduarda Gervini Zampieri Centeno, 2023 winner of the Brain Structure and Function Award
Author: Alexander Holmes
Editors: Elisa Guma, Elizabeth DuPre, Simon Steinkamp
Eduarda tells us about her work developing open-source Python pipelines.
Eduarda Gervini Zampieri Centeno is a Neuroscience PhD candidate at the Bordeaux Neurocampus and a part-time Research Assistant at Amsterdam UMC (Vrije Universiteit Medisch Centrum; VUmc). Passionate about Open Science, her PhD thesis focuses on developing open-source Python pipelines for songbird research and understanding the brain correlates behind vocal learning. At the VUmc, Eduarda promotes and implements Open Science practices within her team and coordinates a working group for departmental transition toward this framework.
Before her PhD, Eduarda completed a 1-year Master's thesis project, applying Python-based topological data analysis to resting-state fMRI datasets from glioma patients. At the end of this intership, she compiled and published her experience as a practical tutorial freely available here. This work culminated in her paper titled “A hands-on tutorial on network and topological neuroscience” and for this work, Eduarda received the Brain Structure and Function award at the 2023 Organization for Human Brain Mapping (OHBM) annual meeting, held in Montreal (amongst other awards [1], [2]).
Eduarda also has experience hosting ReproducibiliTea journal clubs in Bordeaux and Amsterdam; she has organized the first Bordeaux Neurocampus Open Science Workshop, is a member of the Open Science Community Amsterdam board (OSCA), and recently became the Open Science Expert representing the Netherlands at Knowledge Exchange. Read on to hear more from Eduarda!
Q1: What are some advantages to using topological data analysis to standard graph theory methods?
Eduarda Gervini Zampieri Centeno (EGZC): Topological data analysis is a promising approach for studying complex systems and high-dimensional datasets. Compared to traditional methods, it is less susceptible to noise which allows for the computation of abstract data shapes, particularly in biological datasets. This framework enables researchers to explore intrinsic properties in data and go beyond the pairwise relationships traditionally used in graph-theoretical approaches. However, topological data analysis is not simply the opposite of graph theory; instead, it is a complementary approach. Combining these frameworks can yield exciting and relevant results, such as this article by our team. Given that the brain functions in a complex and non-linear way, tools (or combinations of tools!) that enable the investigation of higher order and complexity in data can greatly contribute to the accumulation of knowledge and progress in the field.
Q2: How do you think employing these methods will affect our understanding of human brain mapping?
EGZC: With the advent of new tools, we are constantly presented with new avenues for exploring the functioning and organization of the human brain. In particular, I believe that topological data analysis can better reconstruct and extract valuable information from brain data that has been inaccessible thus far. These new methods should bring us closer to a comprehensive understanding of one of the most complex biological systems.
Q3: What were the greatest challenges you faced when putting together the theoretical overview?
EGZC: Personally, the biggest challenge I faced was during the COVID-19 pandemic, working on this project while living abroad as a master's student. It was a difficult time for everyone, and, some days, it was hard to stay motivated working remotely. Fernando Antônio Nóbrega Santos one of my supervisors—and I had some in-person meetings to discuss this manuscript when possible, but the work was mostly done during lockdowns. However, this also allowed me to immerse myself in the work, put together something I'm very proud of, and learn to not take these opportunities to meet with my team in person for granted.
Professionally, the biggest challenge was understanding how to communicate the knowledge of mathematics and physics into a neuroscientific language. My background is in biology rather than natural sciences, which made communicating this information more difficult. However, this was also why the final material is accessible and comprehensive; as an interdisciplinary team, we understood the struggle and the need to communicate clearly across disciplines very well. I'm genuinely grateful to my supervisors and co-authors, who supported me in creating this material and shared my open science goals from the very beginning. This team experience was a great motivation booster to power through those days and deliver a project that now brings us so much joy.
Q4: How did you first become interested in this topic?
EGZC: This project was born from a mix of interests and interdisciplinary collaboration between Fernando Antônio Nóbrega Santos, Linda Douw, and myself. My primary motivation was to write my master's thesis in the style of a neuroscience Python tutorial that would align with open science and be helpful to others. Fernando, our maths and physics expert, was expanding his horizons toward neuroscience back then. In contrast, Linda, the team leader and clinical neuroimaging specialist, was taking a move toward projects beyond classical graph theoretical metrics and moving her lab to Python. They were excited and willing to teach me network neuroscience and algebraic topology, and in turn, I was eager to learn, code, and share our work with the world. I found our willingness to work together and exchange ideas very exciting. To summarise, the topic became of interest through my great experience working with my team and finding a great place to explore my passions while learning many new things.