Application du rééchantillonnage stochastique de l'échelle en détection-estimation de l'activité cérébrale par IRMf

This paper presents an application of the submission[1] to functional neuroimaging. This joint work aims at simulating less correlated samples in a MCMC algorithm by including a sampling step of a scale parameter when the forward model is bilinear with respect to the unknown parameters. Our application concerns the joint detection-estimation of brain activaty in functional Magnetic Resonance Imaging, where one tries to recover both the shape of the regional Hemodynamic Response Function and stimulus-dependent activation maps. The gain that we obtained is illustrated on realistic simulations.

[1]  Jean-Baptiste Poline,et al.  Dealing with the shortcomings of spatial normalization: Multi‐subject parcellation of fMRI datasets , 2006, Human brain mapping.

[2]  Jérôme Idier,et al.  Spatial Mixture Modelling for the Joint Detection-Estimation of Brain Activity in fMRI , 2007, 2007 IEEE International Conference on Acoustics, Speech and Signal Processing - ICASSP '07.

[3]  J B Poline,et al.  Joint detection-estimation of brain activity in functional MRI: a Multichannel Deconvolution solution , 2005, IEEE Transactions on Signal Processing.

[4]  Jean-Baptiste Poline,et al.  Joint detection-estimation of brain activity in fMRI using an autoregressive noise model , 2006, 3rd IEEE International Symposium on Biomedical Imaging: Nano to Macro, 2006..

[5]  Hichem Snoussi,et al.  Bayesian blind separation of generalized hyperbolic processes in noisy and underdeterminate mixtures , 2006, IEEE Transactions on Signal Processing.