Mathematical Oncology Systems Analysis Imaging Center (MOSAIC)

总结：总体：数学肿瘤学系统分析成像中心 胶质母细胞瘤（GBM）是最具侵袭性的原发性脑癌，是最异质性的癌症之一， 包括肿瘤内和肿瘤间。GBM是肿瘤性胶质瘤细胞和非肿瘤性/反应性胶质瘤细胞的混合物。 脑实质，有助于整个可成像的肿瘤质量。因此，蜂窝内容，包括 细胞密度和细胞组成，对于了解一个 考虑到肿瘤。虽然MRI提供了良好的软组织对比度，可以无创地表征解剖结构， 不存在将对肿瘤的细胞成分的空间和时间理解整合的方法 从体内成像推断。 越来越清楚的是，精确的肿瘤学策略依赖于定量和预测性的 了解每个患者体内癌症复杂系统的状态。我们的最新发现 小组已经揭示了两个关键的机会，我们寻求利用我们提出的数学肿瘤学系统 影像分析中心（MOSAIC）。首先，对我们的GBM图像定位活检队列进行分子分析 已经启发了神经胶质瘤组织状态作为组织样本的复合分类的概念。我们的研究结果 从单核RNAseq揭示了肿瘤和非肿瘤细胞的特定亚群和细胞表型， 肿瘤细胞显示出限制潜在串扰信号传导的不同的共居模式。二是 已经发现GBM活检的临床MRI特征的数学建模和机器学习分析是 能够预测体内GBM生物学的局部区域特征。这些基于图像的模型提供了承诺， 跟踪肿瘤内和肿瘤间异质性的各个方面，这是以前在患者护理过程中无法实现的。 我们的总体中心愿景是建立一个概念框架，以了解组织状态相关的细胞 发生在胶质瘤中的成分转变以及解释MRI相对于这些变化的方法 关键的细胞表型具体来说，在项目1中，我们将探索针对不利（反应迟钝）的策略。 组织状态以导航癌症复合系统向更有利（响应性）组织的转变 states.在项目2中，我们将利用数学建模和机器学习方法来融合MRI和 图像定位活组织检查量化组织状态以使得能够跟踪患者接收中的组织状态变化 护理标准和免疫治疗策略。因此，我们的MOSAIC与CSBC倡议完全一致， 将实验生物学与计算建模相结合，使用成像物理学的方法， 数学肿瘤生长建模，图像引导活检，分子生物学，机器学习， 整合生物信息学，以开发癌症系统生物学的有效进展。

SUMMARY: OVERALL: MATHEMATICAL ONCOLOGY SYSTEMS ANALYSIS IMAGING CENTER Glioblastoma (GBM), the most aggressive primary brain cancer, is amongst the most heterogeneous of cancers, both intra- and inter-tumorally. GBMs are an admixture of neoplastic glioma cells and non-neoplastic / reactive brain parenchyma that contribute to the overall imageable tumor mass. As such, cellular content, including both cellular density and cellular composition, is critically important for understanding the status and evolution of a given tumor. Although MRI provides excellent soft tissue contrast and can noninvasively characterize anatomy, no methods exist to integrate a spatial and temporal understanding of the cellular components of the tumor inferred from imaging in vivo. It has become increasingly clear that precision oncology strategies rely on a quantitative and predictive understanding of the state of the cancer complex system evolving within each patient. Recent findings from our group have revealed two key opportunities we seek to leverage in our proposed Mathematical Oncology Systems Analysis Imaging Center (MOSAIC). First, molecular analysis of a cohort of our image-localized biopsies of GBM have inspired the concept of Glioma Tissue States as a composite classification of tissue samples. Our findings from single nucleus RNAseq reveal that specific subpopulations and cellular phenotypes of neoplastic and non- neoplastic cells show distinct patterns of co-habitation constraining potential cross-talk signaling. Second, we have found mathematical modeling and machine learning analyses of clinical MRI features of GBM biopsies are able to predict loco-regional features of GBM biology in vivo. These image-based models provide the promise to track aspects of intra- and inter-tumoral heterogeneity previously unattainable during patient care. Our overall center vision is to build a conceptual framework to understand tissue state-associated cellular composition transitions that happen in glioma and the ways to interpret MRI relative to those changes for these key cellular phenotypes. Specifically, in Project 1 we will explore strategies to target unfavorable (unresponsive) tissue states to navigate transitions of the cancer complex system towards more favorable (responsive) tissue states. In Project 2 we will leverage mathematical modeling and machine learning approaches to fuse MRI and image-localized biopsy quantified tissue states to enable tracking tissue state changes in patient receiving standard of care and immunotherapy strategies. Thus, our MOSAIC perfectly aligns with the CSBC initiative, integrating experimental biology with computational modeling, using methods from imaging physics, mathematical tumor growth modeling, image-guided biopsies, molecular biology, machine learning, and integrative bioinformatics to develop validated advances in cancer systems biology.