Estimating tissue deformation between functional images induced by intracranial electrode implantation using...

This paper examines a solution to the general problem of accurately relating points within functional data acquired before and after subdural intracranial electrode implantation. We develop an approach based on nonrigid registration of high resolution anatomical MRI acquired together with the functional data. This makes use of a free-form B-Spline deformation model and registration is recovered by maximizing the normalized mutual information between the preimplant MRI and the postimplant MRI. We apply the approach to estimate the tissue deformation induced by the presence of intracranial electrodes over 15 patient studies. Maximum tissue displacements of 4 mm or greater were observed in all cases either in the cortex or around the ventricles due to CSF loss. In studies involving larger 4 x 4 grids, local tissue displacement estimates exceeded 10 mm from the preimplant brain shape. The key issue with this approach is whether the deformation estimates are contaminated by the presence of susceptibility-induced imaging artifacts. We therefore evaluate the deformation estimates in recovering alignment of essentially identical SPECT studies of eight patients acquired before and after electrode placement. An ROI-based analysis of the variance of resulting subtraction values between pre- and postimplant SPECT was carried out in regions of tissue below electrode grids. Results indicate for all cases a substantial reduction in residual SPECT subtraction artifacts to a level comparable to that in an equivalent region of undeformed tissue.