Dynamic bayesian networks: representation, inference and learning
Dynamic bayesian networks: representation, inference and learning
| Hi-index | 0.00 |
In this work we demonstrate improvement of the analysis of functional neuroimaging by combining electromagnetic measurements and functional MRI. We show that magnetoencephalography and functional MRI can complement each other improving estimation of neural activity and BOLD response. Tracking hidden neural activity is performed as inference of latent variables in a dynamic Bayesian network with continuous parameters. Inference is performed using a particle filter. We demonstrate that MEG and fMRI fusion improves estimation of the hidden neural activity and smoothes tracking of the BOLD response. We demonstrate that joint analysis stabilizes the differential system and reduces computational requirements.