High-resolution functional magnetic resonance imaging in the human midbrain

Human midbrain contains nuclei with critical functions for sensation and perception. In particular, the superior colliculus (SC) mediates eye and head movements, the allocation of attention, and multi-sensory integration that are all critical to visually guided behavior. Inferior colliculus (IC) plays a critical in auditory guided behavior. Both of these structures are small (<1 cm diameter) and divided functionally and cytoarchitecturally into multiple laminae. Application of functional magnetic resonance imaging to these nuclei is challenging because, compared to cerebral cortex, they exhibit weaker functional contrast, greater noise, and are positioned deep inside the cranium where signal-to-noise ratio is lower for modern head coils. 

Nevertheless, we have developed methods to resolve functional activity across the surface and within the laminae of the colliculi. Methods include the use of interleaved and multi-echo spiral trajectory pulse sequences using 1.2-mm functional voxels, and surface modeling for visualization that also yields depth coordinates for laminar analysis. We have also evaluated the hemodynamic response function in the colliculi, which are significantly different than those in cortex.

These methods have yielded several interesting confirmations and new findings in the colliculi. In SC, we have demonstrated the detailed polar-angle representation of the visual field and confirmed the precise alignment of this representation for covert visual attention in its most superficial layers. However, when attention is deployed upon a threshold-contrast stimulus, the activity moves deeper, likely toward the intermediate laminae known to mediate eye movements in animal models. In fact, we also confirmed the polar-angle representation for saccadic eye movements within these intermediate layers. Most recently, we have begun experiments to examine somato-visual integration in the deepest layers of SC using a protocol that involves tactile stimulation to perform a visual discrimination task. In IC, we have confirmed a depth-dependent tonotopic representation, but the representation was not evident across the whole of IC, possibly because of variable hemodynamics.