Real-time computer vision tracking of stemness

描述(由申请人提供)：本提案旨在解决对新的干细胞工程工具包的需求，该工具包能够调节成体干细胞的体外扩增(自我更新，无分化)，以便为临床应用创造足够数量的细胞。这项建议的目标是：1)开发一种计算机视觉系统，在相位对比图像中实时跟踪体外细胞分裂的时空历史；2)使用这些数据自动得出对称性、分裂时间、汇合和预测种群增长模型的实时度量；以及3)应用该系统开发新的策略，在生长因子介导的条件下最大化成体干细胞扩增的产量。“实时”是指在图像采集之间的15分钟间隔内“在线”完成处理，这些信息在细胞培养过程中作为过程监测和控制的反馈可用。在目标1中，将使用多模式细胞跟踪模块来实现对密集细胞群体的稳健的图像分割和跟踪，这些模块从不同的角度表示细胞状态和运动并对其进行推理，然后融合和协调它们各自的输出以做出协作决策。为了在基于矢量的图形处理硬件上快速实现，将制定多模式模块。该系统将测量：在发展中的肌源性干细胞(MDSCs)融合群体中培养的每个细胞的增殖、谱系(即亲子关系)、静止和凋亡状态，包括永生化的多能C2C12、原代小鼠MDSCs和原代人类MDSCs。在目标2中，该系统将被应用于有效地发现和表征细胞扩张行为和确定标称培养条件，以利用细胞外基质上固定化生长因子的生物印刷组合阵列来最大化对称性。在目标3中，将使用自适应继代培养策略演示对细胞扩张的实时监测和控制；预测生长模型将发出信号，指示何时暂停细胞培养，然后进行荧光激活的细胞分选，以排除凋亡或分化细胞，从而选择性地丰富传代培养的增殖率最高的细胞。将使用体外成骨和肌肉生成作为范例分化指标来监测不同继代培养的分化潜能的维持。

DESCRIPTION (provided by applicant): This proposal addresses the need for new stem cell engineering toolsets that can regulate adult stem cell expansion (self-renewal, without differentiation) ex vivo in order to create sufficient numbers of cells for clinical applications. The goals of this proposal are to: 1) develop a computer vision system that tracks in real-time the spatiotemporal histories of cell divisions in vitro in phase-contrast imagery; 2) use this data to automatically derive real-time metrics of symmetry, division times, confluence, and predictive population growth models; and, 3) apply this system to develop novel strategies that maximize the yield of adult stem cell expansion in growth factor-mediated conditions. 'Real-time' means that processing is completed 'on-line' in the 15 minute intervals between image acquisitions, and this information is available during cell culture as feedback for process monitoring and control. In Aim 1, robust image segmentation and tracking of dense cell populations will be achieved using multi-modal cell tracking modules that represent and reason about cell states and motion from different perspectives, and then fuse and coordinate their respective outputs to make collaborative decisions. The multi-modal modules will be formulated for fast implementation on vector-based graphics processing hardware. The system will measure: proliferation, lineage (i.e., parent-daughter relationship), quiescence, and apoptotic states of each cell in culture within developing confluent populations of muscle-derived stem cells (MDSCs), including immortalized pluripotent C2C12, primary mouse MDSCs, and primary human MDSCs. In Aim 2, the system will be applied to efficient discovery and characterization of cell expansion behaviors and determination of nominal culture conditions to maximize symmetry using bio-printed combinatorial arrays of immobilized growth factors on extracellular matrix substrates. In Aim 3, real-time monitoring and control of cell expansion will be demonstrated using an adaptive subculturing strategy; the predictive growth model will signal when to suspend cell culture, followed by fluorescent-activated cell sorting to exclude apoptotic or differentiating cells, thus selectively enriching subcultures for cells exhibiting the highest proliferation rates. Maintenance of differentiative potential will be monitored between subcultures using in vitro osteogenesis and myogenesis as paradigm differentiation indices.