Determining the cortical target of transcranial magnetic stimulation.

Determining the cortical region that is effectively targeted by TMS to induce a reproducible behavioral effect is a non-trivial problem. In mapping experiments, a grid of coil positions is used to systematically assess the TMS effect on, e.g. muscle responses or error rates. The center-of-mass (CoM) of the response distribution is projected onto the cortex to determine the likely target site, implicitly assuming the existence of a single, contiguous target. The mapping results, however, often contain several local maxima. These could either stem from measurement noise, or hint towards a distributed target region. Critically, the calculation of a CoM, by design, treats multiple maxima as if they were noise. Here, a stringent hierarchical sigmoidal model fitting approach is developed that determines the cortical target(s) from TMS mapping based on electric field calculations. Monte-Carlo simulations are used to assess the significance and the goodness-of-fit of the sigmoidal fits, and to obtain confidence regions around the calculated targets. The approach was applied to mapping data on visual suppression (N=7). In all subjects, we reliably identified two or three neighboring targets commonly contributing to the suppression effect (average distance+/-SD: 7.7+/-2.3 mm). This demonstrates that (i) the assumption of a single CoM is not generally valid and (ii) the combination of TMS mapping with the fitting approach has a cortical resolution of <1 cm. The estimates for the field strength necessary to achieve 50% of the maximal suppression effect vary noticeably across subjects (mean+/-SD: 139+/-24 V/m), indicating inter-individual differences in the susceptibility to TMS.