Content area
Abstract
Vision is one of the primary senses of humans and other animals. By transforming and decoding visual information, one can perform essential tasks, including environment perception, visual cognition, visual memory, etc. Good visual representation relies on functional organization and mechanisms in the visual system. In highly visual animals, like primates and carnivores, visual information is captured by different photoreceptors in the retina and transformed from optical signal to electrical signal. The electric signal is further relayed to different regions in the visual thalamus, the lateral geniculate nucleus (LGN) by three types of retinal ganglion cells (RGC). Neurons in different LGN regions are projected to their afferent zones in different parts of the primary visual cortex (V1). Retina, LGN, and V1 composed the early visual system. The three distinct processing pipelines from retina to LGN to V1 are also known as parallel streams. As three parallel streams are segregated functionally and anatomically from the retina to V1, neurons in each visual area also showed a cellular topography, that maps from the visual space to its corresponding neurons, known as retinotopy. The parallel streams and retinotopy in the early visual system provided a concise and efficient signal transmission from lower to higher visual area.
The encoded visual information gets more complex as it relays from lower to higher visual area. Neurons in LGN acted like dot detectors: they preferred center-surrounded visual stimulation. Whereas neurons in V1 preferred visual stimulation with different axial orientations, like an edge detector. V1 neurons that have the same orientation preference clustered in columns through cortical layers from the pial surface to white matter. In layers 3 and 4, neurons with different orientation preferences are organized in a pinwheel fashion. However, neurons in the deepest portion of a column, layer 6, don’t always share the same orientation preference as neurons in the more superficial portion of a column. Additionally, the columnar organization in ventral V1 is not investigated as the multi-neuron cellular imaging is not accessible to the back part of the brain. Aside from the feedforward signal transmission, another mechanism, corticogeniculate (CG) feedback, also modulates the visual information in the early visual system. Even though the visual information in LGN is governed by its feedforward input from RGC, CG neurons can also refine the responses of LGN neurons onto which they project, including sharpening the spatial and temporal preference, reducing the response latency, and increasing the information capacity and spiketiming precision. However, the CG feedback properties are mostly studied with electrophysiology with a couple of pairs of LGN and CG neurons in V1. As CG neurons are mostly located in layer 6 of V1, the spatial arrangement of their projected LGN neurons has not been fully investigated. Additionally, the modulatory effect of these CG neurons on their LGN neurons has not been evaluated at a populational level.
In this thesis, we applied ultrasound imaging (fUS) to investigate the functional organization in the ferret visual system. Compared with functional MR, fUS can achieve better spatio-temporal resolution to visualize the functional organization, including retinotopy and orientation preference map, at a lower cost. When compared with extracellular recording, fUS provides wider coverage and better accessibility to all visual areas, especially ventral V1 and different LGN layers. We first imaged the ferret V1 on one hemifield, which is most accessible to most image modalities. By analyzing the fUS activation pattern in 3D, we discovered that fUS can image submillimeter functional architecture in ferret V1. These activation patterns, appearing in both dorsal and ventral V1, correspond to columnar structures with the same orientation preference. As we measured the shape and intervals of these columns, we found the orientation columns appeared to be cone-shaped, where activation near the pial surface has a larger area and activation near white matter is much smaller. The distance between the nearest columns measured by fUS is consistent with previous measurements with optical imaging. Then we studied the relationship between the activation pattern in both V1 and LGN, we demonstrated that fUS can reveal retinotopy, a coarse and universal functional organization, in both V1 and LGN. Additionally, the activated area and volume in V1 and LGN are strongly positive-correlated: Visual stimuli that induce larger activation in LGN always induce larger activation in V1. By imaging the orientation column in V1 and retinotopy in V1 and LGN, we proved that fUS has sufficient resolution to investigate the functional organization and feedforward mechanism in the ferret early visual system. Finally, we turned to the other important mechanism in the visual system: corticogeniculate feedback. By combining fUS with optogenetics, which enabled us to manipulate neuronal activity with external light triggers through genetic engineering, We showed that fUS can identify the optogenetic activation from infected CG neuronal clusters in ferret V1. Additionally, we imaged the populational impact of CG feedback on LGN, which is not accessible with traditional methods like extracellular recording or optical imaging. Together, we showed that fUS has sufficient image resolution and sensitivity in the research of the ferret visual system, which created a solid foundation for fUS application in studying the functional organization and signal relay in other sensory areas with different animal models.
Details
