Distorting anatomy to test MEG models and metrics

The magnetoencephalographic and electroencephalographic (M/EEG) source reconstruction problem is an ill-posed model inversion, so it must be constrained by imposing biologically and physically plausible assumptions. Different M/EEG source reconstruction methods entail different assumptions about the underlying current distribution, yet all produce subjectively plausible current estimates. This work aims to develop an objective method that can be used to test any M/EEG analysis pathway. We make use of advances in diffeomorphic brain shape modeling to construct a set of parametrically deformable cortical surfaces that are representative of the population. These deformed (surrogate) brains provide a quantifiable parametric distortion from the ground-truth anatomy. If the current flow giving rise to the MEG data were generated on the true cortical manifold, MEG current estimates should be selective of the true anatomy. We show in simulation and with empirical data how the correct reconstruction assumptions depend closely on the true anatomy. We present a method to quantify the performance of MEG source reconstruction algorithms (and metrics of fit) in terms of millimeters of distortion.