Joerg Magerkurth1,
Laura Mancini1, William Penny2, Guillaume Flandin2,
John Ashburner2, Caroline Micallef1, Enrico De Vita1,
Tarek A. Yousry1, John Thornton1, Nikolaus Weiskopf2
1Department
for Brain Repair and Rehabilitation, UCL Institute of Neurology, London,
United Kingdom; 2Wellcome Trust Centre for Neuroimaging,
University College London, London, United Kingdom
Intra-operative fMRI combined with pre-surgical fMRI planning could provide crucial information to guide neurosurgery. The intra-operative (1.5T) MR scanner at our institution presents lower signal-to-noise and contrast-to-noise ratios and significantly larger distortions than the 3T MR scanner used pre-operatively. We present a pre-processing and Bayesian analysis method for pre- and intra-operative fMRI. While classical inference identifies only activated area, the Bayesian approach labels three areas: activated, non-activated, and areas where the data do not allow for a robust classification. The method was tested on 10 healthy volunteers in a passive movement paradigm, which could also be used under anesthesia.