AI-driven biomarker analysis of intact whole brains imaged at micron and sub-micron resolution

以微米和亚微米分辨率成像的完整全脑的人工智能驱动生物标志物分析

• 批准号: 10330017
• 负责人: Katherine Cora Ames
• 金额: $ 22.06万
• 依托单位: LIFECANVAS TECHNOLOGIES, INC.
• 依托单位国家: 美国
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-02-01 至 2023-01-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10330017
• 关键词: 3-Dimensional; Adoption; Algorithms; Amyloid beta-Protein; Antibodies; Artificial Intelligence; Astrocytes; Automobile Driving; Back; Binding Proteins; Biological Markers; Brain; Brain imaging; Cell Density; Cell membrane; Cells; Collaborations; Communities; Complex; Computer software; Cytoplasm; Data; Data Analyses; Data Set; Detection; Development; Disease; Feedback; Future; Generations; Glial Fibrillary Acidic Protein; Health; Heterogeneity; Image; Image Analysis; Imaging Techniques; Immunohistochemistry; Individual; Institutes; Label; Libraries; Location; Maps; Microscopy; Modeling; Morphology; Mus; Neurologic; Neurons; Neurosciences; Nuclear; Organ; Organ Preservation; Output; Pattern; Positioning Attribute; Process; Proteins; Proteomics; Quality Control; Research Contracts; Resolution; Rosaniline Dyes; Technology; Testing; Three-Dimensional Image; Three-Dimensional Imaging; Three-dimensional analysis; Tissue imaging; Tissues; Training; algorithm development; artificial intelligence algorithm; base; biomarker-driven; cell type; cellular imaging; computerized data processing; cost; design; extracellular; improved; insight; neuronal cell body; new technology; programs; relating to nervous system; segmentation algorithm; software development; statistics; submicron; terabyte; tissue processing; tool; user friendly software; user-friendly

Abstract. Whole-organ 3D immunohistochemistry is revolutionizing the field of neuroscience, enabling unprecedented insight into the distribution of neural cells and neurological markers throughout the brain in health and disease. LifeCanvas Technologies is at the forefront of the new field of spatial proteomics, providing a complete workflow for whole-organ preservation, tissue clearing, immunohistochemical labeling, and imaging. Nevertheless, an ongoing challenge for such studies is the need to rapidly, reproducibly and rigorously quantify terabyte-sized datasets from whole-organ imaging efforts. While progress has been made in applying Artificial Intelligence (AI) tools to enable detection of cellular and sub-cellular markers in neural tissue, one-size-fits-all algorithms are inadequate for analyzing complex, information-rich brain datasets due to varying biomolecular expression patterns (e.g. nuclear, cytoplasmic, membrane-bound) and region-specific heterogeneities in cell density and neural cell types. However, AI-driven algorithms targeting a subset of labeling patterns can be effective provided the availability of adequate training data. LifeCanvas Technologies LCT is optimally positioned to develop highly accurate algorithms serving a wide range of detection tasks through its access to high volumes of whole-organ image data containing a variety of label expression patterns via its Contract Research Organization and user base. LCT proposes to develop a data analysis program, SmartAnalytics, which will embed a suite of AI algorithms within a user-friendly software package to identify labeled cell locations and characterize morphological features across the whole brain at cellular and sub-cellular resolution. Specifically, LCT will use intact, 3D immunolabeled mouse brains to design AI algorithms to detect labeled cells imaged at cellular resolution and generate further algorithms for the segmentation of labeled features imaged at sub-micron resolution. Data from LCT’s Contract Research Organization and academic collaborations will be continually fed back to improve and expand the library of detection algorithms available within SmartAnalytics, and these developments will drive further customer adoption and enhancement of future versions of the software. SmartAnalytics will guide users through model application, quality-control testing, and the generation of output products such as figures and summary statistics. In summary, SmartAnalytics will be an evolving and user- friendly workflow execution program that enables neuroscientists to take full advantage of their 3D image data, driving new discoveries in brain function, development and disease.

抽象的。全器官3D免疫组织化学正在彻底改变神经科学领域， 前所未有的深入了解神经细胞和神经标记物在整个大脑中的分布， 和疾病LifeCanvas Technologies处于空间蛋白质组学新领域的最前沿， 完整的工作流程，用于整个器官保存、组织清除、免疫组织化学标记和成像。 然而，这些研究的一个持续挑战是需要快速，可重复和严格地量化 整个器官成像的TB级数据集。虽然在应用人工智能方面取得了进展， 智能（AI）工具，能够检测神经组织中的细胞和亚细胞标记物，一刀切 算法不足以分析复杂的、信息丰富的大脑数据集， 表达模式（如核、胞质、膜结合）和细胞内区域特异性异质性 密度和神经细胞类型。然而，针对标记模式子集的AI驱动算法可以被 有效的前提是提供充分的培训数据。LifeCanvas Technologies LCT处于最佳位置 开发高度精确的算法，通过其访问大量的检测任务， 包含各种标签表达模式的整个器官图像数据， 组织和用户基础。LCT建议开发一个数据分析程序SmartAnalytics，该程序将 在用户友好的软件包中嵌入一套人工智能算法，以识别标记的细胞位置， 以细胞和亚细胞分辨率表征整个大脑的形态特征。具体地说， LCT将使用完整的3D免疫标记小鼠大脑来设计AI算法，以检测成像的标记细胞， 细胞分辨率，并生成进一步的算法，用于亚微米成像的标记特征的分割 分辨率来自LCT合同研究组织和学术合作的数据将持续提供 返回以改进和扩展SmartAnalytics中可用的检测算法库， 这些发展将推动客户进一步采用和改进未来版本的软件。 SmartAnalytics将指导用户完成模型应用、质量控制测试和输出生成 产品，如数字和汇总统计。总之，SmartAnalytics将是一个不断发展和用户- 友好的工作流程执行程序，使神经科学家能够充分利用他们的3D图像数据， 推动大脑功能、发育和疾病的新发现。