An automated tool that trains a multimodal classification model (EEG + fMRI) to predict phenotypic variables (sex, age, diagnosis) and assess whether fusing both modalities improves prediction compared to each modality alone.
This project investigates how resting-state functional connectivity differs between sleep and wake states using simultaneous EEG and fMRI data from the OpenNeuro “Simultaneous EEG and fMRI signals during sleep from humans” dataset. By comparing default mode network connectivity and EEG frequency-band connectivity across states, the project found minimal differences at the network level but a reliable increase in theta-band connectivity during sleep, highlighting the value of combining modalities to characterize state-dependent brain connectivity.
This project aimed to further my intuitive understanding of fMRI data. Around 20 interactive/static figures of various statistics of raw fMRI data, confounds and atlased data were produced. Special efforts have been made to make the analysis highly and easily reproducible.
This project aimed to develop a BIDS App for performing Connectome-Wide Association Studies (CWAS) on fMRI connectivity matrices. The result is a GitHub repository that can be installed via pip, enabling analyses on BIDS-formatted connectomes. Integration tests ensure the pipeline runs reliably, and a dedicated website provides full documentation and example outputs.
This project explores how different fMRI analysis methods reveal distinct aspects of the neural representation of fear, including a mass univariate approach (GLM), a machine learning (decoding) approach, and representational similarity analysis (RSA) approach. While GLM identified some expected activation patterns and machine learning failed to decode fear ratings reliably, RSA revealed modest but significant structure in frontal regions, highlighting the value of methodological triangulation in cognitive neuroscience.
This study examines dorsolateral prefrontal cortex (dlPFC) and posterior parietal cortex (PPC) connectivity and dlPFC BOLD time series in ADHD versus typically developing (TD) children during the cued stop-signal task (CSST) using fMRI data from OpenNeuro ds005899. It is hypothesised that stronger dlPFC-PPC connectivity will be found in the ADHD group.
This project aims to extract and analyze fMRI signals from the hippocampus during spatial navigation, using a reproducible workflow based on open tools and data formats.
This mini-project for the 2025 Brainhack School is part of my PhD dissertation on Late-Life Cognitive Heterogeneity, where I examine the neural correlates of cognitive dispersion – a measure of within-individual variability – using neuropsychological and fMRI data from the Midnight Scan Club dataset (OpenNeuro ds000224).
This project aimed to explore tools and techniques used to analyze fMRI brain activation at the first level using data from an open-access dataset. We produced a comprehensive Jupyter Notebook that provides a step-by-step guide to running the analyses, including applying the GLM, defining contrasts, and generating various brain activation maps.
This project aims to examine how the brain supports two types of inference—active and passive—during the process of posterior belief integration. I applied trial-by-trial analysis to track participants’ learning trajectories over time.
Copyright (c) 2026, BrainHack School; all rights reserved.
Template by Bootstrapious. Ported to Hugo by DevCows.