Improving large-scale 3D EM data alignment for connectomics of sensory deprivation

Abstract

Does the environment play a crucial role in shaping synaptic connectivity and neuronal circuit organization? The rodent somatosensory cortex serves as a well-established model for studying experience-dependent plasticity, particularly through sensory deprivation paradigms such as whisker trimming. While numerous studies have examined structural and functional changes in deprived barrels, findings remain inconsistent, highlighting the need for nanoscale-resolution, dense reconstructions to resolve these debates. Electron microscopy (EM)-based connectomics enables high-resolution imaging of neuronal circuits, offering the potential for unprecedented insights into deprivation-induced network effects. Extending such reconstructions to cortical columns presents major computational challenges, particularly in the accurate alignment of electron micrographs. Misalignments, cracks, and jumps in 3D EM volumes impede automated analysis and tracing of neurites, limiting high-throughput, petabyte-scale data processing. In this thesis, we analyze large cortical layer-specific EM volumes of 1741824 μm3 in both spared and deprived barrel columns. By examining synaptic connectivity in layers II, III, and IV, we identify over 700,000 synapses per volume, with a significantly higher synapse density in deprived barrels compared to spared ones. To enhance the accuracy of large-scale reconstructions, we develop an improved volume alignment pipeline that minimizes distortions and misalignments. This methodological advancement should facilitate robust comparative analyses of sensory deprivation across multiple cortical columns, paving the way for more comprehensive studies of experience-dependent plasticity.