Fast Event-Related Mapping of Population Fingertip Tuning Properties in Human Sensorimotor Cortex at 7T

Abstract

fMRI studies that investigate somatotopic tactile representations in the human cortex typically use either block or phase-encoded stimulation designs. Event-related (ER) designs allow for more flexible and unpredictable stimulation sequences than the other methods, but they are less efficient. Here, we compared an efficiency-optimized fast ER design (2.8-s average intertrial interval; ITI) to a conventional slow ER design (8-s average ITI) for mapping voxelwise fingertip tactile tuning properties in the sensorimotor cortex of six participants at 7 Tesla. The fast ER design yielded more reliable responses compared with the slow ER design, but with otherwise similar tuning properties. Concatenating the fast and slow ER data, we demonstrate in each individual brain the existence of two separate somatotopically-organized tactile representations of the fingertips, one in the primary somatosensory cortex (S1) on the postcentral gyrus, and the other shared across the motor and premotor cortices on the precentral gyrus. In both S1 and motor representations, fingertip selectivity decreased progressively, from narrowly-tuned Brodmann area (BA) 3b and BA4a, respectively, toward associative parietal and frontal regions that responded equally to all fingertips, suggesting increasing information integration along these two pathways. In addition, fingertip selectivity in S1 decreased from the cortical representation of the thumb to that of the pinky.

Keywords: event-related; motor cortex; population receptive field; somatosensory cortex; tactile perception; ultra-high field fMRI.

Figure 1.

Fingertip stimulation sequence for the three experimental designs. In all three designs, stimuli were 400 ms, 50-Hz vibrotactile stimulations at the tip of one of the five fingers of the left hand (corresponding to the five colors in this figure). In the phase-encoded design (A), each fingertip was stimulated for 4 s (ITI = 100 ms) in clockwise (thumb to pinky, shown in this figure) or counter-clockwise order (not shown). In the slow and fast ER designs, each event consisted of two stimulations separated by 100 ms. Events were presented in random order, with an ITI of 4 to 12 s in the slow ER design (B) or 2 s in the fast ER design (C). In addition, the fast ER sequence included two null events for every five fingertip stimulations, resulting in an average ITI of 2.8 s.

Figure 2.

Fingertip-specific cortical representations in somatotopically-organized regions, as revealed by the phase-encoded design in participants 1–3 (see Extended Data Fig. 4-1A for all participants). Colored voxels show the phase of the sinewave best fitting the phase-encoded timeseries, with the correspondence between phase and fingertip preference shown in the color scale on the right. Phase maps were thresholded at a coherence value corresponding to p < 0.05, FDR-corrected. All participants showed two somatotopically-organized regions, one on the postcentral gyrus and the other, smaller, on the precentral gyrus (encircled with dotted outline). Some participants also showed hints of somatotopic responses in the central sulcus (BA3a and BA4p), but we cannot exclude that these were because of mis-registrations or extra-vascular BOLD response (see Extended Data Fig. 2-1). Fingertip-specific cortical representation on the postcentral and precentral gyri (colored outlines) were delineated by binning the phase values as indicated in the color scale and were used later on to define ROIs. Dotted black lines represent the likely location of borders between BAs, as derived from a probabilistic cytoarchitectonic atlas.

Figure 3.

Comparison of response estimates and somatotopic maps between fast and slow ER designs. A, GLM-estimated HRF averaged across the five fingertip-specific S1 cortical representations defined in Figure 2, in response to the preferred and adjacent fingertips of each representation. Error bars represent the standard error (SE) of the voxelwise GLM estimates at each time point, averaged across voxels, cortical representations, and participants. Estimate SEs were smaller for the fast than the slow ER designs. B, Single-parameter response estimates for the preferred and adjacent fingertips, derived using a GLM analysis using the average participant-specific preferred-fingertip HRF as a canonical HRF. Again, parameter estimate SEs were smaller in the fast than the slow ER design. C, Composite fingertip preference maps for each ER design in the first three participants. Each of the five colors represents voxels responding significantly more to a given fingertip than to the other four. Each fingertip map is thresholded at p < 0.05 FDR-corrected. Colored outlines represent the postcentral and precentral fingertip-specific cortical representations from the phase-encoded localizer. The number of significant voxels was much larger in the fast than the slow ER data.

Figure 4.

ER-derived single-participant maps of fingertip preference and selectivity in participants 1–3 (see Extended Data Fig. 4-1 for all participants). A, Composite fingertip preference maps showing which voxels respond more to a given fingertip stimulation than to the other fingertips on average, thresholded at p < 0.05, FDR-corrected for each fingertip contrast. This type of map highlights the same somatotopically-organized regions as the phase-encoded maps in Figure 2. B, Composite activation maps showing which voxels respond significantly to the stimulation of each fingertip compared with baseline, thresholded at p < 0.05, FDR-corrected for each fingertip contrast. This map highlights voxels that responded significantly either mostly to one fingertip (same fingertip colors as panel A), to two (usually adjacent) fingertips (blend of two fingertip colors) or more than two fingertips (white and off-white hues). C, Fingertip pRF center parameter maps computed by fitting voxelwise tuning curves with Gaussians, thresholded at p < 0.05 FDR-corrected according to the main effect of fingertip in the GLM analysis. This map highlights the same voxels as the phase-encoded (Fig. 2) and ER-derived composite preference maps (panel A). D, pRF width parameter maps (from the same fit as C), thresholded at p < 0.05 FDR-corrected according to an F test testing for significant positive activation across any of the five fingertips in the GLM analysis. Extended Data Figure 4-2 shows the maps obtained after including right-hand button presses as a covariate in the GLM (see text for details).

Figure 5.

Voxelwise tuning curves and tuning widths in all somatotopically ordered and nonordered ROIs, averaged across voxels in each ROI and across all participants. A, ROI-average voxelwise tuning curves with best-fitting Gaussian and associated Gaussian FWHM. Error bars represent the SE of each ROI-average response estimate across participants. ROI-average voxelwise tuning curves were obtained by centering voxelwise tuning curves on each voxel’s preferred fingertip (derived from the phase-encoding data) before averaging across voxels (for details, see Materials and Methods). Extended Data Figure 5-1 illustrates the extent of the circularity bias obtained when using ER-derived instead of phase-encoding-derived preferred fingertip estimates. B, ROI-average pRF width parameter estimate, measuring fingertip tuning width. Error bars represent the SE of ROI-averaged pRF width estimates across participants. Horizontal bars indicate statistical significance for post hoc pairwise comparisons, corrected for all possible pairwise comparisons. Both panels show that fingertip tuning width was smallest in BA3b and increased going posteriorly on the postcentral gyrus and going anteriorly in the precentral gyrus.

Figure 6.

Comparison of pRF parameters obtained from the slow and fast ER designs A, pRF center maps for participants 1–3, derived from the slow and fast ER data, respectively (see Extended Data Fig. 6-1 for all participants). As in Figure 4C, maps are thresholded according to the main effect of fingertip in the GLM analysis (from the respective slow or fast ER dataset). More voxels were significantly fingertip specific in the fast than the slow ER design. For voxels that were significant in both designs, the preferred fingertips were generally identical. B, pRF width maps derived from the slow and fast ER data, respectively. As in Figure 4D, maps are thresholded according to an F test testing for significant positive activation across any of the five fingertips in the GLM analysis (from the respective slow or fast ER dataset). Significantly activated voxels were similar between the two designs. In somatotopically-organized regions, pRF width estimates were mostly similar between the two designs (except for participant 3 who showed narrower tuning in the fast than the slow design). In BA2 and posterior regions, and in BA6, pRF width estimates differed between the two designs in a way that varied between participants: participants 1 and 3 showed wider pRFs in the fast than the slow design, whereas participant 2 showed the opposite pattern. C, ROI-average voxelwise pRF width derived separately from the slow and fast ER designs (data averaged across all participants). Despite (nonsignificant) differences in pRF width between the two ER designs in some ROIs (see text), the same spatial pattern was observed in both designs, with narrower fingertip tuning in more primary cortical areas (BA3b, BA1, and BA4a) increasing toward secondary somatosensory (BA2 and posterior) and premotor areas (BA6). Error bars as in Figure 5.

Figure 7.

Voxelwise tuning curves and tuning widths in fingertip-specific ROIs in each somatotopically-organized BA. Panels A, B are as in Figure 5, but for fingertip-specific regions divided by both fingertip stimulation preference and BA. Tuning width increased from the thumb to the pinky in BA3b and BA1, but not in BA4a/6. BA3b was also more narrowly tuned than either BA1 or BA4a/6. Error bars as in Figures 5-6.

Figure 8.

Cortical magnification functions in ordered BA ROIs. The cortical magnification functions were estimated by computing the geodesic cortical distances between consecutive pairs of adjacent fingertip ROIs in each BA. Error bars as in Figures 5-7. There was no significant evidence for cortical magnification of the radial vs ulnar fingers in either of the three BAs (but the index-thumb distance was larger than the other distances in BA3b and BA4a/6). There was cortical magnification in postcentral compared with precentral ordered regions, in that distances were about twice as large in BA3b and BA1 than in the BA4a/6.