| Paper: | MLSP-P2.8 | ||
| Session: | Bioinformatics and Biomedical Applications | ||
| Time: | Wednesday, May 19, 13:00 - 15:00 | ||
| Presentation: | Poster | ||
| Topic: | Machine Learning for Signal Processing: Biomedical Applications and Neural Engineering | ||
| Title: | SEMI-BLIND DECONVOLUTION OF NEURAL IMPULSE RESPONSE in fMRI USING A GIBBS SAMPLING METHOD | ||
| Authors: | Salima Makni; Commissariat à l'Énergie Atomique | ||
| Philippe Ciuciu; Commissariat à l'Énergie Atomique | |||
| Jérôme Idier; IRCCyN (CNRS) | |||
| Jean-Baptiste Poline; SHFJ/CEA | |||
| Abstract: | In functional Magnetic Resonance Imaging (fMRI), the Hemodynamic Response Function (HRF) represents the impulse response of the neurovascular system. Its identification is essential for a better understanding of cerebral activation since it provides a typical time course of the subject response to a stimulus. In [Ciuciu03], the authors have developed a voxel-based HRF estimation method. Here, we propose an extension that takes the spatial homogeneity of the HRF into account. Our goal is to characterize any specified region of interest (ROI) by a single impulse response. Since the measured signal amplitude may strongly vary across voxels, a voxel-dependent neural response level has to be simultaneaoulsy estimated for each type of stimulus. We are thus faced to a semi-blind deconvolution inverse problem since time arrivals of the neural response are known:they correspond to times of stimulation. To cope with this issue, we introduce specific prior information about the HRF and the neural response. Finally, we develop a MCMC approach to approximate the posterior mean estimates of unknown quantities. Realistic simulation results show the improvement brought by our formulation. | ||
| Back | |||
Home -||-
Organizing Committee -||-
Technical Committee -||-
Technical Program -||-
Plenaries
Paper Submission -||-
Special Sessions -||-
ITT -||-
Paper Review -||-
Exhibits -||-
Tutorials
Information -||-
Registration -||-
Travel Insurance -||-
Housing -||-
Workshops