C03 Predictive mechanical tumor markers: solid stress and multiscale viscoelastic data analysis

癌症的力学特征包括异常高的固体应力以及内在力学组织特性的剧烈和空间异质性变化。虽然固体应力引发机械感觉信号促进肿瘤进展，但机械异质性和组织流动性有利于细胞疏解和转移扩散。这个子项目的总体目标是确定对肿瘤形成敏感的机械参数变化的预测指纹，并对肿瘤的侵袭性增殖和非干扰转变具有特异性。为此，C03将首先开发量化体内固体应力的工具，最终将其与组织流动性（co2）和机械异质性（C01）相结合，用于肿瘤的力学分析。肿瘤生长产生的固体应力形成了一个有利于肿瘤进展的生物物理环境，必须在体内分析，因为它在切除后不存在。因此，C03的目的是利用训练好的神经网络结合断层弹性成像获得的体内刚度图，从高分辨率3D MRI的大染色分析中获得的体内变形场中获得固体应力。固体应力的新标记将支持多重力学数据的协调，因为已知固体应力会改变肿瘤中由于压缩硬化而产生的表观刚度。为了解决机械参数协调问题，C03将编译所有子项目的数据库，包括来自细胞、类器官、组织以及体内患者的机械参数。考虑固体应力、组织流动性和长度尺度上的机械异质性的指导性学习方法将用于指定与本研究单元中获得的所有机械数据兼容的最佳拟合粘弹性模型和参数范围。模态和频率无关的机械参数将与临床、组织学和多组学数据结合使用无监督学习，以便识别关于离体和体内系统的肿瘤生物学的高灵敏度、特异性和预测性的集群。总的来说，我们的工作计划建立在主要假设的基础上，即肿瘤获得的癌症机械特征和MRE的无创量化决定了肿瘤的侵袭性、恶性潜能和治疗反应。为了验证这一假设，C03将专注于(i)结合其他子项目提供的其他癌症特异性机械标记（如流动性和异质性）的固体应力量化，（ii）学习模态总体和组织固有本构参数，以及（iii）多频MRE和定量MRI诊断价值的自动评估。总的来说，C03旨在将所有a - b - c项目中产生的数据转化为充分反映肿瘤侵袭性潜力及其对治疗反应性的机械概况路线图。

The mechanical traits of cancer include abnormally high solid stress as well as drastic and spatially heterogeneous changes in intrinsic mechanical tissue properties. Whereas solid stress elicits mechanosensory signals promoting tumor progression, mechanical heterogeneity and tissue fluidity are conducive to cell unjamming and metastatic spread. The overall aim of this subproject is to identify predictive fingerprints of mechanical parameter changes that are sensitive to tumor formation and specific to aggressive proliferation and unjamming transitions in tumors. To this end, C03 will first develop the tools to quantify solid stress in vivo, which will ultimately be combined with tissue fluidity (C02) and mechanical heterogeneity (C01) for mechanical profiling of tumors. Solid stress generated by growing tumors forms a biophysical environment in favor of tumor progression that must be analyzed in vivo, as it is not present after resection. Therefore, C03 aims to derive solid stress from in-vivo deformation fields obtained from large stain analysis of high-resolution 3D MRI using trained neuronal networks in combination with in-vivo stiffness maps acquired by tomoelastography. The novel marker of solid stress will support the reconciliation of multiplexed mechanical data, as solid stress is known to alter the apparent stiffness in tumors due to compression stiffening. To address mechanical parameter reconciliation, C03 will compile a library of data from all subprojects that include mechanical parameters from cells, organoids, tissues as well as in-vivo patients. Guided learning approaches that account for solid stress, tissue fluidity and mechanical heterogeneity across length scales will be used to specify best-fitting viscoelastic models and parameter ranges compatible with all mechanical data obtained in this research unit. Modality- and frequency independent mechanical parameters will be combined with clinical, histological and multi-omics data using unsupervised learning in order to identify clusters of high sensitivity, specificity and predictivity with respect to tumor biology of ex vivo and in vivo systems. Overall, our work plan builds on the main hypothesis that the mechanical hallmarks of cancer acquired by a tumor and noninvasively quantified with MRE determine tumor aggressiveness, malignant potential and treatment response. To verify this hypothesis, C03 will focus on (i) solid stress quantification in combination with other cancer-specific mechanical markers such as fluidity and heterogeneity provided by the other subprojects, (ii) learned modality-overarching and tissue-intrinsic constitutive parameters, and (iii) automated assessment of the diagnostic value of multifrequency MRE and quantitative MRI. Collectively, C03 aims to translate the data generated in all A-B-C projects into a roadmap of mechanical profiles which fully reflect the aggressive potential of a tumor and its responsiveness to treatment.