Astrocytes are the most abundant glial cells in the central nervous system.They perform a diverse array of functions,with a critical role in structural integrity,synapse formation,and neurotransmission.These cells exh...Astrocytes are the most abundant glial cells in the central nervous system.They perform a diverse array of functions,with a critical role in structural integrity,synapse formation,and neurotransmission.These cells exhibit substantial regional heterogeneity and display variable responses to different neurological diseases.Such diversity in astrocyte morphology and function is essential for understanding both normal brain function and the underlying mechanisms of neurological disorders.To investigate this heterogeneity,we developed a novel method for the selective and sparse labeling of astrocytes in various brain regions.This technique utilizes a dual adeno-associated virus system that allows for the expression of Cre recombinase and enhanced green fluorescent protein under the control of the glial fibrillary acidic protein(GfaABC1D)promoter.The system was tested in C57BL/6J mice and successfully labeled astrocytes across multiple brain regions.The method enabled the detailed visualization of individual astrocytes-including their intricate peripheral processes-through three-dimensional reconstructions from confocal microscopy images.Furthermore,the labeling efficiency of this dual adeno-associated virus technology was validated by examining astrocyte function in a spared nerve injury model and through chemogenetic modulation.This innovative approach holds great promise for future research because it enables a more comprehensive understanding of astrocyte variation not only in spared nerve injury but also in a broad spectrum of neurological diseases.The ability to selectively label and study astrocytes in different brain regions provides a powerful tool for exploring the complexities of these essential cells and their roles in physiological and pathological conditions.展开更多
Seismic facies identification is a crucial link in seismic data interpretation. Deep learning technology can enhance the efficiency and accuracy of automatic seismic facies identification. However, deep learning metho...Seismic facies identification is a crucial link in seismic data interpretation. Deep learning technology can enhance the efficiency and accuracy of automatic seismic facies identification. However, deep learning methods typically rely on large amounts of labeled data, and in practical applications, the labeling cost of seismic data is high, with great difficulty. Additionally, basic well logging data cannot be directly utilized. To this end, this paper proposes a semi-supervised automatic seismic facies identification method based on ultra-sparse well logging labels. First, based on the HRNet, a seismic facies identification model that uses onedimensional well logging labels is built for supervision. Second, to preserve the vertical characteristics of seismic data, this paper develops a sparse label sampling module(SLSM) that conducts samples around the well logging labels without slicing the seismic data vertically, thus retaining its vertical depth features and laying a solid foundation for subsequent semi-supervised learning tasks. Third, in terms of the lateral correlation of seismic data, the region growing training strategy(RGTS) is proposed, which expands the information from well logging labels to the entire seismic volume through an iterative growing process. Experiments on real-world data show that the proposed model achieves a mean intersection over union(MIo U) of 79.64% by using only 32 one-dimensional well logging labels, which account for less than 0.5% of the total data volume. This approach provides references for conducting seismic facies identification in areas with sparse and locally distributed well logging data, demonstrating promising application potential.展开更多
基金National Natural Science Foundation of China,No.32271148(to JW)the National Key Research and the Development Program of China,No.2023M740625(to ML)+1 种基金the Natural Science Foundation of Guangdong Province,Nos.2021B1515120050(to HW)and 2023A1515110782(to ML)and Key R&D Program of Ningxia Hui Autonomous Region,No.2024BEG02027(to JW).
文摘Astrocytes are the most abundant glial cells in the central nervous system.They perform a diverse array of functions,with a critical role in structural integrity,synapse formation,and neurotransmission.These cells exhibit substantial regional heterogeneity and display variable responses to different neurological diseases.Such diversity in astrocyte morphology and function is essential for understanding both normal brain function and the underlying mechanisms of neurological disorders.To investigate this heterogeneity,we developed a novel method for the selective and sparse labeling of astrocytes in various brain regions.This technique utilizes a dual adeno-associated virus system that allows for the expression of Cre recombinase and enhanced green fluorescent protein under the control of the glial fibrillary acidic protein(GfaABC1D)promoter.The system was tested in C57BL/6J mice and successfully labeled astrocytes across multiple brain regions.The method enabled the detailed visualization of individual astrocytes-including their intricate peripheral processes-through three-dimensional reconstructions from confocal microscopy images.Furthermore,the labeling efficiency of this dual adeno-associated virus technology was validated by examining astrocyte function in a spared nerve injury model and through chemogenetic modulation.This innovative approach holds great promise for future research because it enables a more comprehensive understanding of astrocyte variation not only in spared nerve injury but also in a broad spectrum of neurological diseases.The ability to selectively label and study astrocytes in different brain regions provides a powerful tool for exploring the complexities of these essential cells and their roles in physiological and pathological conditions.
文摘Seismic facies identification is a crucial link in seismic data interpretation. Deep learning technology can enhance the efficiency and accuracy of automatic seismic facies identification. However, deep learning methods typically rely on large amounts of labeled data, and in practical applications, the labeling cost of seismic data is high, with great difficulty. Additionally, basic well logging data cannot be directly utilized. To this end, this paper proposes a semi-supervised automatic seismic facies identification method based on ultra-sparse well logging labels. First, based on the HRNet, a seismic facies identification model that uses onedimensional well logging labels is built for supervision. Second, to preserve the vertical characteristics of seismic data, this paper develops a sparse label sampling module(SLSM) that conducts samples around the well logging labels without slicing the seismic data vertically, thus retaining its vertical depth features and laying a solid foundation for subsequent semi-supervised learning tasks. Third, in terms of the lateral correlation of seismic data, the region growing training strategy(RGTS) is proposed, which expands the information from well logging labels to the entire seismic volume through an iterative growing process. Experiments on real-world data show that the proposed model achieves a mean intersection over union(MIo U) of 79.64% by using only 32 one-dimensional well logging labels, which account for less than 0.5% of the total data volume. This approach provides references for conducting seismic facies identification in areas with sparse and locally distributed well logging data, demonstrating promising application potential.
