Via multi-modality MRI neuroimages, network neuroscience aims to record, analyze, and model the brain regions and their interactions from a network and integrative perspective. Meanwhile, deep learning enables the computational model to automatically learn abstract and efficient representations from data and has achieved great advances in medicine and biology. The convergence of network neuroscience and deep learning will undoubtedly emit a lot of sparks of thought. However, challenges exist in several aspects including “the data sparsity fact in network neuroscience” versus “the data-hungry nature of deep learning”, “the explicit knowledge requirement from network neuroscience” versus “the unsatisfactory explainability of deep learning”.
This Research Topic proposes to tackle some of the challenges in the convergence of network neuroscience and deep learning. There are at least three objectives: to generate novel deep learning methods for the construction, representation, and understanding of the multiscale and dynamic brain network; to discover new spatiotemporal knowledge about the brain at the levels of region, connection, community, sub-network, and network; to apply the converged methods in precise management of neurological diseases. We welcome high-quality Original Research articles or Review Articles related to the novel methods, discoveries, and applications in the convergence of network neuroscience and deep learning.
Potential areas of interest include, but are not limited to:
• Parcellation of brain regions
• Deep learning for multimodality MRI registration
• Denoising of functional MRI signal
• Normalization of fMRI from different scanners
• Estimation of brain connectivity
• Generation or synthesis of MRI signal and brain network
• Graph convolutional network in neuroscience
• Causality analysis
• Few-shot learning
• Deep learning explainability
• Identification of network biomarkers
• Predictive models of various clinical outcomes
Via multi-modality MRI neuroimages, network neuroscience aims to record, analyze, and model the brain regions and their interactions from a network and integrative perspective. Meanwhile, deep learning enables the computational model to automatically learn abstract and efficient representations from data and has achieved great advances in medicine and biology. The convergence of network neuroscience and deep learning will undoubtedly emit a lot of sparks of thought. However, challenges exist in several aspects including “the data sparsity fact in network neuroscience” versus “the data-hungry nature of deep learning”, “the explicit knowledge requirement from network neuroscience” versus “the unsatisfactory explainability of deep learning”.
This Research Topic proposes to tackle some of the challenges in the convergence of network neuroscience and deep learning. There are at least three objectives: to generate novel deep learning methods for the construction, representation, and understanding of the multiscale and dynamic brain network; to discover new spatiotemporal knowledge about the brain at the levels of region, connection, community, sub-network, and network; to apply the converged methods in precise management of neurological diseases. We welcome high-quality Original Research articles or Review Articles related to the novel methods, discoveries, and applications in the convergence of network neuroscience and deep learning.
Potential areas of interest include, but are not limited to:
• Parcellation of brain regions
• Deep learning for multimodality MRI registration
• Denoising of functional MRI signal
• Normalization of fMRI from different scanners
• Estimation of brain connectivity
• Generation or synthesis of MRI signal and brain network
• Graph convolutional network in neuroscience
• Causality analysis
• Few-shot learning
• Deep learning explainability
• Identification of network biomarkers
• Predictive models of various clinical outcomes
