Spatial resolution of fMRI in the human parasylvian cortex: Comparison of somatosensory and auditory activation

Abstract

In spite of its outstanding spatial resolution, the biological resolution of functional MRI may be worse because it depends on the vascular architecture of the brain. Here, we compared the activation patterns of the secondary somatosensory and parietal ventral cortex (SII/PV) with that of the primary auditory cortex and adjacent areas (AI/AII). These two brain regions are located immediately adjacent to each other on opposite banks of the Sylvian fissure, and are anatomically and functionally distinct. In 12 healthy subjects, SII/PV was activated by pneumatic tactile stimuli applied to the index finger (0.5 cm² contact area, 4 bar pressure), and AI/AII by amplitude-modulated tones (800 Hz carrier frequency, modulated at 24–36 Hz). Functional images were obtained with a 1.5-T scanner and were evaluated using SPM99. Sensitivity of fMRI activation in this unselected sample was 71% for tactile and 83% for auditory stimulation. Group analysis showed activation of SII/PV by tactile and activation of three locations in AI/AII by auditory stimuli. Distributions extended to the opposite side of the fissure (19–58% after tactile and 13–14% after auditory stimulation, depending on the side of stimulation/hemisphere). Morphometry of individual sulcal anatomy revealed that the course of the Sylvian fissure varied by 5.3 mm (SD) in vertical direction. Taking this into account, SII/PV was located 5.8 ± 2.7 mm above the Sylvian fissure, whereas AI/AII was located 6.3 ± 1.7 mm below the Sylvian fissure. Even in individual analysis, the most significant voxel after tactile stimuli in one subject was found on the “wrong” side of the fissure; this error could be ascribed to the spatial normalization procedure. These data show that fMRI signals may overlap substantially, even if the activated regions are separated by 12 mm across a major sulcus. Spatial normalization to an atlas template can introduce additional variance. Individual sulcal anatomy should be preferred over mean atlas locations.

Introduction

Functional magnetic resonance imaging (fMRI) is a valuable tool to localize cortical activation in a wide variety of stimulation paradigms. Although technical developments allow a precision ranging from millimeters to submillimeters (e.g., Menon and Goodyear, 1999), the biological spatial resolution of fMRI is limited by the vascular architecture of the brain. Moreover, the neural correspondence, measured as correlation between cortical activity and the evoked BOLD response, is becoming less than 50% for voxel sizes below 2.8 mm³ (Ugurbil et al., 2003). At present, it is unclear, to what extent potential overlaps in blood supply and spatial filtering may compromise the interpretation of fMRI findings in humans. In the central region, for example, it was found that sensory stimuli coactivate the precentral motor cortex (MI), and that motor paradigms coactivate the postcentral somatosensory cortex (SI, Puce et al., 1995, Spiegel et al., 1999, Yetkin et al., 1996). This coactivation pattern is plausible because there are overlapping anatomical connections (ascending input to MI, descending corticospinal tract originates partly in SI, corticocortical interconnections) and there is other functional evidence for interactions between sensory and motor areas (Cohen and Starr, 1987, Forss and Jousmäki, 1998, Gandevia and Burke, 1990, Knecht et al., 1993, Rushton et al., 1981). For the same reason, however, the region around the central sulcus cannot be used to assess the effective spatial resolution of fMRI.

In order to determine the spatial resolution of fMRI, we selected another region of the brain: the parasylvian cortex. The parasylvian cortex is structurally similar to the perirolandic region with a large fissure–the Sylvian fissure or lateral sulcus–separating the temporal lobe from the frontal and parietal lobes. In contrast to the central sulcus, the Sylvian fissure separates functionally distinct structures: the primary and secondary auditory cortices (AI, AII) in the temporal lobe, and the secondary somatosensory cortex and the parietal ventral area (SII/PV) in the parietal lobe. Therefore, it is possible to judge whether the focus of fMRI activation is in the correct brain area.

The primary auditory cortex (AI) is located below the Sylvian fissure on the superior temporal gyrus. According to studies using cortical recordings in humans, AI is confined to the medio-dorsal third of Heschl's gyrus, (Celesia, 1976, Howard et al., 2000, Liegeois-Chauvel et al., 1991). A recent study using probabilistic mapping of 27 post mortem brains revealed a high variability of the size of the portion of the cytoarchitectonically determined AI (Te 1 in that study) in relation to Heschl's gyrus, covering 16–92% of it (Rademacher et al., 2001). Functionally, AI is involved in the processing of speech and pure tones (Hari et al., 1984, Pantev et al., 1989, Morosan et al., 2001). Recent neuroimaging studies and subdural recordings in humans revealed multiple secondary auditory cortical areas around AI (Di Salle et al., 2001, Rivier and Clarke, 1997) which could be associated to higher order functions of auditory processing like spatial discrimination, tone duration, and complex sounds (Engelien et al., 2000). For these areas, we will use the collective term “AII” throughout the paper.

The secondary somatosensory cortex (SII) is situated within the parietal operculum in the upper bank of the Sylvian fissure (Burton et al., 1993). SII and the adjacent parietal ventral area (PV) contain multiple somatotopic representations of the body surface and are surrounded by several other somatosensory areas within the frontoparietal operculum and the adjacent insula (Burton et al., 1993, Craig and Dostrovsky, 1997, Disbrow et al., 2000, Krubitzer et al., 1995). As part of the somatosensory tactile system, SII/PV are involved in the processing of mechanical stimuli and seem to play a major role in tactile object recognition (Caselli, 1993) and in processes of allocated attention (Timmermann et al., 2001). The SII/PV region and adjacent insula also play a role in the processing of noxious stimuli and pain sensation (Peyron et al., 2002, Schnitzler and Ploner, 2000, Treede et al., 2000, Vogel et al., 2003).

In this study, we investigated the biological spatial resolution of fMRI using stimuli evoking cortical activation which a priori can be assigned to auditory (AI, AII) or somatosensory (SII/PV) brain regions. For activation of SII/PV, we used tactile stimulation of the fingers. For the activation of AI, we used amplitude modulated pure tones. Both stimuli were previously shown to activate these regions in electrophysiological studies in humans (Hoechstetter et al., 2001, Gutschalk et al., 1999, Schneider et al., 2002).