Improve reproducibility and transparency in the field of neuroimaging by applying nonparametricstatistical methods and writing R packages.

Brain data analyses involves many steps and every step is prone to errors and uncertainties. Ignoring uncertainties can potentially leading to overconfident conclusions. To improve reproducibility it is important to propagate errors throughout the analysis. One crucial step in functional imaging studies is image registration to align subject-specific brain anatomy to a common brain atlas. This step ensures that we compare anatomically similar brain regions. During my exchange at EPFL hosted by Professor Van De Ville, we will develop software that integrates registration uncertainty into a functional brain analysis workflow.