Figure 2.

Structural and Functional Localization of Stimulation Sites on the Medial Frontal Wall and Associated Behavior

In separate panels for each patient are shown renders of the MR structural, MR functional, and CT post-electrode-implantation imaging, all non-linearly transformed into standard MNI stereotactic space by a unified normalization and segmentation procedure implemented in SPM12. For each patient, the MR structural image (a pre-implantation T1-weighted 0.94 × 0.94 × 1.1 mm acquisition from which the MNI normalization parameters were derived for all other imaging) is represented for clarity by the estimated gray matter compartment only, with isolines corresponding to the 90%, 80%, and 70% probability contours, in that order of increasing intensity, cut through a parasagittal plane at x = −4 mm. The functional imaging data, performed before implantation and derived from blocked verb repetition (yellow) or verb generation (red) compared with rest, were used to compute SPM t-statistic maps of significant task-related activation, which were then rigidly co-registered to the structural scan via the mean echoplanar image and subsequently transformed into MNI space. Semi-transparent contours of the clusters on the medial wall are thresholded at p = 0.05 family-wise error corrected, except for verb generation in DH where weak activation necessitated a drop in threshold to p = 0.001 uncorrected. The CT post-electrode-implantation image, a 0.43 × 0.43 × 1.2 mm acquisition, was rigidly co-registered to the pre-operative MR volume and then non-linearly adjusted by a unified normalization and segmentation procedure with the previously estimated, smoothed native space MR tissue compartments applied as priors. The non-linear adjustment was applied to compensate for the subtle but noticeable descent of the dorsal surface following craniotomy so as to improve the accuracy of contact localization in the dorsoventral plane. As with the others, this adjusted image was then transformed into MNI space using the parameters derived from the MR image, resampled to 1 × 1 × 1 mm resolution. Each grid contact was then visualized by rendering with a contour thresholded at metal density, within a region of interest enclosing the medial wall so as to exclude both bone and contacts elsewhere in the brain. The critical loci where a behavioral effect was observed are enclosed by dashed ellipses, lying on the ventral border of the pre-supplementary motor area. Note that since the stimulation current was biphasic, the polarity of the electrodes reversed at 50 Hz.

The insets show violin plots of the distributions of the reciprocals of the inter-movement intervals—essentially instantaneous frequency, in Hz—for the alternating tasks the patients performed, both manually and vocally, while the critical contacts were stimulated. The manual task consisted of self-paced, repetitive finger flexion and extension movements; the vocal task consisted of equally self-paced, repetitive single syllable vocalizations of the form “la-la-la.” The red lines index the change in the locations of the distributions, showing a significant increase in behavioral frequency in the manual task for DH (p < 0.001, Bonferroni adjusted, marked ^∗∗∗) and in the vocal task for LW (p = 0.030, Bonferroni adjusted, marked ^∗), consistent with effector-specific inhibition of procrastination. See also Figures S1–S3, Tables S1 and S2, and Movie S1.