Anibal Sólon Heinsfeld

This project provides tools for neuroscience researchers to analyze their neuroimaging data. In the last year, I've had the opportunity to improve its code and develop new tools to coexist with it.

This tool aims to generate brain parcellations – divisions of the brain based on its functioning – for individuals and for a group of individuals. The BASC method applies bootstrapping (bagging) aggregation, improving the parcellation reproducibility. Reproducibility is important given that the human brain shows anatomical and functional variability between participants. Dr. Nikolaidis and I worked together to bring this tool to be easily applied by other researchers, and provide statistical proof of this method for several neuroimaging datasets. The publication for this tool is on its way!

As part of the project Feita Na Biblioteca (Made at the Library), I helped in the development of the backend and a GraphQL API for the Descartae app. This app aims to help Brazilian users to locate where to discard waste properly. We build it up using Node.JS and MongoDB, and since everything is containerized using Docker, I already migrated from different Kubernetes clusters seamlessly. This project was developed in conjunction with Caravan Studios (https://www.caravanstudios.org).

A simple tool for researchers to find matching subjects. It is common in clinical studies to have more healthy participants than participants from within a condition. This tool helps the researcher to select healthy participants by keeping the distribution of data descriptors, such as age and sex. For each condition participant, the algorithm selects a healthy participant by minimizing the Gower distance of their descriptors. The Gower distance is ideal for this task since data descriptors can handle numerical and categorical data. In the end, the researcher will have an unbiased selection of participants for her study, improving its statistical validity.