Mattias Villani

This paper introduces a Bayesian model for fMRI analysis that uses Gaussian process priors to estimate brain activity and hemodynamics without assuming a linear time-invariant system.

This paper introduces anatomically informed Bayesian spatial models for fMRI analysis that improve posterior probability maps by incorporating anatomical structure.

The paper introduces a Bayesian non-central chi regression model for fMRI and DWI data, showing improved accuracy in localizing brain activity and estimating diffusion metrics compared to Gaussian models.

The paper introduces a Bayesian heteroscedastic GLM for fMRI data to better handle motion spikes, showing it detects more brain activity by modeling time-varying variance.