cuBNM
cuBNM is a Python package to efficiently simulate brain network models (BNMs) and fit them to empirical neuroimaging data through integrated optimization algorithms. The toolbox can simulate BOLD signal, derive functional connectivity (FC) and functional connectivity dynamics (FCD) from it, and compare them to FC and FCD matrices derived from empirical BOLD signals to assess the goodness-of-fit. These simulations can be used either to explore how parameter variations influence network dynamics, in silico, or to optimize parameters to best match empirical functional data from individual subjects or groups. Such individualized or group-specific models can be used to study hidden simulation-derived features.
While the toolbox can run simulations on both CPUs and GPUs, it is optimized for GPU parallelization to enable massive scaling of the simulations, gaining speed-ups in the order of several hundred to thousand times faster than CPUs. User can chose between a Python an a command-line interface.
Find out more on GitHub or in the Documentation.
Research Software Engineers and Contact Points
Amin Saberi
PhD Student
- Institute of Neurosciences and Medicine (INM)
- Brain and Behaviour (INM-7)
Room 3038