Medial-to-lateral somatotopy is a well-established feature of the human primary somatosensory cortex (SI); however, it is unknown whether, similarly to non-human primates, a rostral-to-caudal somatotopic arrangement exists as well. Therefore, in this functional magnetic resonance imaging (fMRI) study on eight healthy human subjects, five circumscribed skin areas sequentially located on the third finger and the palm of the hand were stimulated with innocuous electrical pulses. Within area 3b of contralateral SI, successive cortical representation sites ordered in a rostral-to-caudal fashion were seen in the group analysis and in six individual subjects. The fingertip was located most rostrally, whereas the proximal parts of the finger as well as the distal palm were represented at more caudal locations. Within area 1, the group analysis revealed a similar pattern of discrete representations. However, in contrast to area 3b, the fingertip was located most caudally, whereas the more proximal parts of the finger were found to be represented rostrally within area 1. Thus, the representation pattern of area 1 appeared as a ‘mirror image’ of that of area 3b. In comparison to the representations of the finger and the distal palm, the proximal palm was found to be represented at a more medial position of the postcentral gyrus.
In primates, the processing of somatosensory information engages, amongst other areas, primary (SI) and secondary (SII) somatosensory cortex, in which the contralateral body surface is represented somatotopically. For human SI, a somatotopic representation of the contralateral body surface in a medial- to-lateral direction along the postcentral gyrus (‘homunculus’) has been found by Penfield and co-workers (Penfield and Boldrey, 1937; Penfield and Rasmussen, 1950; Penfield and Jasper, 1954) using electrical stimulation of the cortical surface of the postcentral gyrus in awake patients during neurosurgery. A very similar representation pattern was also described by Woolsey and colleagues for several mammals measuring evoked potentials from the exposed cortex (Woolsey et al., 1942; Woolsey and Fairman, 1946). Cytoarchitectonically, SI consists of four different areas that are arranged from rostral to caudal and are termed areas 3a, 3b, 1 and 2, according to the classification of Brodmann and Vogt and Vogt, respectively (Brodmann, 1909; Vogt and Vogt, 1919). Microelectrode recordings in non-human primates revealed that the contralateral body surface is represented in each of these four areas (Merzenich et al., 1978; Kaas et al., 1979; Nelson et al., 1980; Sur et al., 1980b; Kaas, 1983; Pons et al., 1985). Evidence for different functional impacts of the subdivisions has been given, demonstrating a particular sensitivity of neurons in areas 3a and 2 to stimulation of deep receptors, whereas neurons in areas 3b and 1 exhibit a predominant responsiveness to stimulation of cutaneous receptors (Powell and Mountcastle, 1959; Iwamura et al., 1993). Furthermore, a hierarchy in sensory information processing was postulated, according to which area 3b can be regarded as ‘SI proper’ (Kaas, 1983). For humans, recent neuroimaging and electrophysiological studies confirmed non-invasively the homuncular (i.e. medial-to-lateral) somatotopic organization of SI (Fox et al., 1987; Baumgartner et al., 1991, 1993; Hari et al., 1993; Gelnar et al., 1998; Kurth et al., 1998; Maldjian et al., 1999; Stippich et al., 1999). Furthermore, there is evidence that in humans there are also multiple representations of a certain body region, i.e. representations within two or more of the subdivisions of SI (Allison et al., 1989; Lin et al., 1996; Burton et al., 1997; Kurth et al., 2000; Moore et al., 2000; Ploner et al., 2000) and that a medial-to-lateral somatotopic representation exists within at least some of these subdivisions (Gelnar et al., 1998; Francis et al., 2000; Kurth et al., 2000; Ruben et al., 2001; Deuchert et al., 2002).
In addition to this medial-to-lateral somatotopic organization, studies in non-human primates demonstrated a rostral-to-caudal-somatotopic arrangement of body regions within these cytoarchitectonic areas (Paul et al., 1972; Merzenich et al., 1978; Kaas et al., 1979). For the hand region of area 3b, the representation of the fingertip has been shown to be located most rostrally, whereas more proximal parts of the finger are represented at slightly more caudal positions (the terms ‘rostral’ and ‘caudal’ always referred to a flattened cortex in these studies; we maintain this classification here too). In human subjects, there are so far only two studies that have explicitly addressed the possible rostral-to-caudal somatotopic organization in area 3b of SI (Hashimoto et al., 1999a,b). In these studies using magnetencephalography (MEG), no clear evidence was found of a possible rostral-to-caudal somatotopic arrangement within this area. This finding may be explained by the limits of spatial resolution in MEG. Taking advantage of the superior spatial resolution of functional magnetic resonance imaging (fMRI), we performed electrical stimulation of five adjacent skin areas located along the distal-to-proximal axis on the third finger and the palm of the right hand in order to address the following questions.
Is there a rostral-to-caudal somatotopic arrangement within area 3b of humans?
If there is, does such a rostral-to-caudal somatotopic arrangement also exist within adjacent areas 1 and 2?
If so, are these rostral-to-caudal somatotopic representations within adjacent subdivisions of SI arranged mirror-reversed as suggested by studies in non-human primates?
Material and Methods
Eight right-handed volunteers (seven male, one female; mean age 30 years, range 25–39 years) without any history of neurological or psychiatric disorders were investigated. The study was approved by the local ethics committee and written consent was obtained from each subject prior to investigation.
Somatosensory stimulation consisted of innocuous electrical stimuli generated by a clinical constant current neurostimulator device (Neuropack 2, Nihon Kohden, Tokyo, Japan). Monophasic square-wave current pulses (frequency 7 Hz, single pulse duration 0.2 ms) were delivered to five electrodes located in-line on the third finger and the palm of the subject’s right hand. Electrodes consisted of a central anode (tip diameter 4 mm) and a concentric cathode (width 2 mm, diameter 12 mm) in order to obtain focal stimulation of a circumscribed skin area. The electrodes were attached to the glabrous skin at the following positions: distal (P1), middle (P2) and proximal phalanx (P3) of the third finger and over the caput (P4) and base (P5) of the third metacarpal bone (Fig. 1). The amplitude for each stimulation site was individually adjusted to 1 mA below threshold intensity for pain sensation. The mean stimulation intensities were 12.9 mA (P1), 11.1 mA (P2), 10.6 mA (P3), 10.7 mA (P4) and 14.0 mA (P5).
fMRI Data Acquisition
MR imaging was performed on a 1.5 T clinical scanner (Magnetom Vision, Siemens, Erlangen, Germany) using a surface coil (CP Flex) for the functional measurements and a head coil for acquisition of the anatomical data sets. The centre of the surface coil was placed over the left parietal cortex approximately at the C3′ position according to the extended 10/20 EEG system. The subject’s head was immobilized by means of a vacuum pad and fixation tape in order to minimize movement-related artefacts. BOLD (blood oxygenation level dependent) sensitive (Bandettini et al., 1992; Frahm et al., 1992; Kwong et al., 1992; Ogawa et al., 1992) echoplanar images (TR = 3 s, TE = 66 ms, flip angle = 90°) were acquired consisting of eight slices (FOV = 256 mm, matrix = 128 × 128, voxel size = 2 × 2 × 3 mm, gap = 0.75 mm, interleaved order of slice acquisition) that were orientated in parallel to the bicommisural plane. A manual shim in the region of interest (postcentral gyrus) was performed to improve the homogeneity of the static magnetic field. For every subject, two successive runs with 420 images each were performed. Within a single run, each site (P1–P5) was stimulated alternately in a blocked order (30 s rest, 30 s stimulation) four times ending with a final rest period (30 s). The initial 10 images were discarded due to T1 saturation effects. Stimulations were applied in a sequential order (P1, P2, P3, P4 and P5) in both runs and since there was no additional task involved, intrinsic attentional effects were rather negligible. For anatomical co-registration a FLASH (fast low angle shot) sequence (TR = 20 ms, TE = 5 ms, flip angle = 30°, 180 slices, voxel size = 1 × 1 × 1 mm) covering the whole brain was applied.
Functional images were motion-corrected (sinc interpolation) with SPM99 (Wellcome Department of Cognitive Neurology, London, UK). Subsequent data analysis was performed using BrainVoyager® 4.2 (BrainVision, Maastricht, The Netherlands). Functional images were aligned to the anatomical data set and both were transformed into Talairach space. First, the data sets were rotated in stereotactic space after the anterior commissure (AC) and the posterior commissure (PC) as well as the midsagittal plane had been determined manually. Then the borders of the cerebrum were specified and together with the AC and PC coordinates the 3D volumes were finally scaled to Talairach space using a piecewise affine and continuous transformation for each of the 12 resulting subvolumes (Talairach and Tournoux, 1988). Functional image data was interpolated to 1 mm isovoxel size and, after linear trend removal, spatially and temporally smoothed with a Gaussian filter (FWHM = 4 mm and FWHM = 2 s, respectively). For computation of statistical maps, multiple regression analysis was employed using the general linear model. A boxcar function (shift of onset = 6 s) was used as a regressor for each of the five stimulation conditions (P1–P5). T-maps of each condition were computed for the individual subjects and the group analysis (n = 8). The thresholds for the statistical maps were set to P < 10−5 uncorrected for multiple comparisons due to a strong a priori hypothesis for the expected activations.
The hand area of SI was determined as the region just posterior to the hand area of the primary motor cortex, which can be identified on axial slices through the brain due to its omega- or epsilon-like shape (Yousry et al., 1997). In order to account for the activation of the different cytoarchitectonic subdivisions of SI, the postcentral gyrus was investigated for discrete activation foci. As cytoarchitectonically defined areas cannot be identified precisely on anatomical landmarks alone (Geyer et al., 1999), for the assignment of the activation foci to these subdivisions an operational definition was used according to a set of recent fMRI studies (Gelnar et al., 1998; Francis et al., 2000; Moore et al., 2000): (i) the fundus of the central sulcus corresponds to area 3a; (ii) the posterior bank of the central sulcus to area 3b; (iii) the crown of the postcentral gyrus to area 1; and (iv) the anterior bank of the postcentral sulcus to area 2. The statistical maps were visualized superimposed onto the individual reconstructed brain surfaces, this being achieved by segmenting and tessellating the grey/white matter boundary and inflating the resulting surface mesh (Linden et al., 1999). The statistical maps of the group analysis were also visualized onto an individual brain surface; in order to exclude that the results were based on the specific individual neuroanatomy of that brain, the T-maps were visualized onto different brains.
Innocuous electrical stimulation of circumscribed areas on the middle finger and the palm of the right hand consistently led to statistically significant signal intensity changes within the hand area of the contralateral postcentral gyrus. In general, stimulation of a particular site on the hand was associated with multiple activation foci across the postcentral gyrus, reflecting the activation of different subdivisions of SI. The activations occurred within areas 3b, 1 and 2, whereas no activations were found within area 3a.
Within area 3b, activations due to stimulation of adjacent sites on the hand were seen at slightly different locations in the rostral-to-caudal direction. The general finding was that the fingertip (P1) was represented most rostrally, i.e. most deeply within the posterior bank of the central sulcus, whereas the representations of the more proximal parts of the finger (P2, P3) and of the distal palm (P4) were located at more caudal positions. Results of the group analysis are given in Figure 1. From a dorsal view, successive activations associated with stimulations of P1, P2, P3 and P4 can be seen within the posterior bank of the central sulcus, which corresponds to area 3b. The most rostral activation occurred due to stimulation of P1. The representation site of P2 was seen caudally adjacent to that of P1 and is followed by that of P3. Activation due to stimulation of P4 was located most caudally. From a lateral view, activation sites on the crown of the postcentral gyrus can also be seen in Figure 1. At this anatomical location, which most probably corresponds to area 1, a pattern of discrete representations was again seen. Caudal to the single P4 activation (which is also visible in Fig. 1, upper part), an activation due to stimulation of P3 is seen. This activation is followed caudally by a second P2 activation. The most caudal activation occurs due to stimulation of P1. In comparison to the closely grouped activations along the rostral- to-caudal direction in response to stimulation of P1, P2, P3 and P4, stimulation of P5 led to an activation at a more medial position within the posterior bank of the central sulcus (area 3b). The Talairach coordinates of the activations of the group analysis are listed in Table 1.
In the individual subjects, a representation pattern similar to that of the group analysis was observed within area 3b. The main finding was that the fingertip (P1) was found to be represented most rostrally, whereas the representations of the more proximal parts of the finger (P2, P3) were located more caudally. Such a representation pattern was seen in six out of the eight subjects. The P3 representation was generally found at a slightly more caudal position as compared to the P2 representation. Signal intensity changes due to stimulation of P4 were seen in three subjects. Activations in response to stimulation of P2, P3 and P4 exhibited a greater extent of overlap as compared to activations due to stimulation of P1 and P2. The results of an individual subject (subject No. 4) are shown in Figure 2. The complete pattern of discrete representations within area 1, as observed in the group analysis, could not be described within the single subjects. The reason for this was that frequently the stimulations did not lead to statistically significant signal intensity changes within area 1, preventing inferences from being made about a discrete representation pattern in this area for the single subjects. Stimulation of P5 was associated with statistically significant signal intensity changes in four out of eight subjects. The P5 representation sites were found at a more medial position as compared to the finger representations. Table 2 gives the Talairach coordinates of the activations occurring within area 3b in the single subjects.
Within area 2, regions of activation corresponding to the different stimulation sites overlapped significantly within the individual subjects, as well as the group; thus, for area 2 it was not possible to establish a somatotopic map.
In this study, we found several discrete statistically significant signal intensity changes across the contralateral postcentral gyrus in response to focal stimulation of a single circumscribed skin area on the right hand. Most probably this reflects the functional activation of different cytoarchitectonic subdivisions of SI, as it is known from microelectrode studies in non-human primates that a specific body region is represented within each of these subdivisions (Merzenich et al., 1978; Kaas et al., 1979; Nelson et al., 1980; Sur et al., 1980b; Pons et al., 1985). For humans, recent neuroimaging studies indicate that this principle of organization can be assumed as well (Lin et al., 1996; Burton et al., 1997; Gelnar et al., 1998; Kurth et al., 1998, 2000; Maldjian et al., 1999; Francis et al., 2000; Moore et al., 2000; Krause et al., 2001; Ruben et al., 2001). Furthermore, previous imaging studies have demonstrated the medial-to-lateral somatotopic representation of single fingers in SI and there is evidence that this somatotopic arrangement can be resolved, at least for area 3b and area 1 (Gelnar et al., 1998; Francis et al., 2000; Kurth et al., 2000; Ruben et al., 2001).
In contrast to the extensive investigations concerning the medial-to-lateral somatotopic organization of human SI, less attention has been given to its potential rostral-to-caudal somatotopy. In non-human primates, microelectrode recordings demonstrated a rostral-to-caudal somatotopic arrangement of body areas within the different cytoarchitectonic subdivisions of SI (Paul et al., 1972; Merzenich et al., 1978; Kaas et al., 1979). Within area 3b, a single finger has been found to be represented in a rostral-to-caudal sequence, starting with the fingertip most rostrally followed by the adjacent skin areas of the middle and proximal phalanx. Our data are in agreement with these findings. The fingertip (P1) was represented most rostrally within the depth of the posterior bank of the central sulcus, whereas the representations of P2 and P3 were found to be located at successively more caudal positions. This finding of a somatotopic arrangement within area 3b is in contrast to results of recent MEG studies by Hashimoto and colleagues using a 122-channel whole-head planar gradiometer system (Hashimoto et al., 1999a) and a 37-channel first-order axial gradiometer system (Hashimoto et al., 1999b). In these studies, a vibratory stimulus of 200 Hz was used to stimulate several circumscribed areas, from the tip of the index finger most distally to the palm of the hand most proximally. In both studies, dipole localization revealed no statistically significant differences in the locations of the dipoles with regard to the peripheral site of stimulation. However, whereas in one study a trend was reported for the finger tip to be located most deeply within the posterior bank of the central sulcus (Hashimoto et al., 1999a), in the second study a reversed order of representations was found (Hashimoto et al., 1999b). The authors mainly explained their findings either by a blurring of the rostral-to-caudal somatotopic organization in humans as compared to non-human primates or by a lack of spatial resolution in their MEG approach. In order to overcome the latter problem, in this study we performed fMRI using a surface coil that allowed us to measure functional data with an improved in-plane resolution of 2 × 2 mm at an increased signal-to-noise ratio. We assume that the higher spatial resolution of our fMRI approach is the main factor that enabled us to demonstrate the rostral-to-caudal somatotopic arrangement within human area 3b, although there were also differences concerning the stimulation paradigm (i.e. electrical versus vibratory stimulation).
Within area 1 of non-human primates, discrete representations of distal and proximal parts of a single finger have also been described (Paul et al., 1972; Merzenich et al., 1978; Kaas et al., 1979). The representations of the distal and proximal parts of the finger in area 1 were found to be arranged in a reversed order in comparison to area 3b, i.e. proximal parts of the finger were found to be represented rostrally, whereas the fingertip was located caudally (Merzenich et al., 1978; Kaas et al., 1979). In our study, evidence for discrete representations of proximal and distal parts of a single finger within area 1 can be inferred from the group analysis. On the crown of the postcentral gyrus, caudally to the P4 activation, a discrete representation pattern in the order P3, P2 and P1 can be seen in a rostral-to-caudal direction. The solitary P4 activation on the anterior crown of the postcentral gyrus might be considered to be located at the border between area 3b and area 1 and, therefore, to reflect the activation of both areas. Compared to area 3b, the representation pattern within area 1 appears to be mirror-reversed and the P4 activation may be regarded as the inflection point. Thus, our data suggest the unexpected curiosity that the arrangement in humans is similar to that in owl monkeys, for which the inflection point between areas 3b and 1 has also been reported to be at the distal palm (Merzenich et al., 1978). In contrast, it seems to differ from the situation described in macaques, where the inflection point was at the proximal phalanx or base of the fingers (Nelson et al., 1980).
For area 2 of non-human primates, there is evidence for a representation pattern similar to that of area 3b, i.e. the fingertip was found to be located rostrally and the proximal parts of the finger were represented caudally (Pons et al., 1985). Extensive overlap of the activations within area 2 in our data prevents any conclusions regarding a somatotopic arrangement within human area 2. A less distinct somatotopy in area 2 is in agreement with previous reports. Whereas overlap between activation sites has been reported for each of the subdivisions of SI, the extent of overlap is different with regard to a particular cytoarchitectonic area. In general, the extent of overlap steadily increases from area 3b to area 2. This finding may be explained by differences in the receptive field characteristics. Microelectrode mapping studies have demonstrated that in the postcentral finger region of alert monkeys, neurons in area 1 and area 2 tended to have larger and more complex receptive fields than in area 3b (Hyvarinen and Poranen, 1978; Sur et al., 1980a). Similarly, Iwamura and colleagues reported a continuous increase in the number of neurons with receptive fields covering multiple fingers or finger and palm from area 3b to area 2 and, as a consequence, the representations of fingers and skin areas on the palm were found to be rather overlapping within area 1 and 2 (Iwamura et al., 1980, 1983a,Iwamura et al., b). In agreement with these invasively recorded data, in a previous fMRI study in human subjects it was shown that the overlap between finger representation sites increased from area 3b to area 1 and 2 (Krause et al., 2001).
The most proximal point of stimulation over the base of the metacarpal bone (P5) was found to be represented medially to the rostral-to-caudal axis of the finger (P1, P2, P3) and distal palm (P4) representations. Microelectrode mapping studies in non-human primates have demonstrated a representation for the radial part of the glabrous hand at a location laterally to the representation of the thumb, whereas the representation of the ulnar part was found medially to the representation of the fifth finger (Merzenich et al., 1978; Kaas et al., 1979; Nelson et al., 1980; Pons et al., 1985). Thus, one explanation for our finding may be that the stimulation site of P5 in our study excited neurons with receptive fields belonging to the ulnar part of the hand.
The data of the group analysis show the feasibility of mapping the rostral-to-caudal somatotopic arrangement within areas 3b and 1 by means of fMRI. This study, furthermore, represents a non-invasive functional approach to determine the border between these anatomically defined areas. As some inter-individual variability has been reported concerning the exact locations between the cytoarchitectonically defined sub-divisions of SI (Geyer et al., 1999), further improvements in imaging technique (for example, by employing higher field strengths) might offer the possibility of determining functionally the borders of the subdivisions, even in individual subjects. In analogy to the visual system, a delineation in this manner may thus serve as basis for properly characterizing the functional impact of each somatosensory subdivision non-invasively in humans.
In summary, this study gives evidence for the existence of a rostral-to-caudal somatotopic organization of human area 3b. A rostral-to-caudal somatotopic arrangement was also observed within area 1, whereas in area 2 an overlapping representation pattern predominated. Representations within areas 3b and 1 were mirror images of each other.
The authors thank Gabriel Curio for providing the skin electrodes and Birol Taskin as well as Thomas Krause for helpful comments on the manuscript. The work was in part supported by the Deutsche Forschungsgemeinschaft (DFG).
Address correspondence to Felix Blankenburg, Department of Neurology, Charité, Humboldt-University, Schumannstrasse 20/21, 10117 Berlin, Germany. Email: email@example.com.
|aIn response to stimulation of P4, a single activation was seen located on the border between areas 3b and 1.|
|aIn response to stimulation of P4, a single activation was seen located on the border between areas 3b and 1.|
|*Statistically significant different from each other (P < 0.05, Wilcoxon signed rank test).|
|*Statistically significant different from each other (P < 0.05, Wilcoxon signed rank test).|