AI enhanced lifetime-based mesoscopic in vivo imaging of tissue molecular heterogeneity

摘要 药物靶向结合的量化被认为是药物开发中最关键的参数 流水线，因为它是治疗行动的中心。然而，这种参数只能通过有创的方法进行评估 体外组织的生化和免疫组织化学(IHC)方法。在此，我们建议整合和 优化可提供直接纵向(多时间点)的多模式光学成像平台 药物-靶向结合分布在相同组织体积上的测量及其与药物的相关性 传递疗效参数，包括肿瘤血管和新陈代谢等药物反应指标。 该成像平台将在人类乳腺肿瘤和活体动物的患者来源的异种移植中得到验证 接受HER2-曲妥珠单抗治疗。此外，由于MFMT是一种间接成像技术，依赖于 在复杂的计算任务上，我们将进一步率先使用深度学习方法来实现FAST、 精确、无参数且用户友好的2D和3D MFMT图像形成。

Abstract Quantification of drug-target engagement is recognized as the most crucial parameter in the drug development pipeline as it is central to therapeutic action. Though, such parameter can only be assessed via invasive biochemical and immunohistochemical (IHC) approaches in ex vivo tissues. Herein, we propose to integrate and optimize a multimodal optical imaging platform that can provide direct longitudinal (multiple time points) measurements of the drug-target engagement distribution across the same tissue volume in correlation with drug delivery efficacy parameters, including, tumor vasculature, and indicators of drug response, such as metabolism. The imaging platform will be validated in human breast tumor and patient derived xenografts in live animals subjected to HER2-trastuzumab therapy. Additionally, as MFMT is an indirect image formation technique relying on complex computational tasks, we will further pioneer the use of Deep Learning methodologies for fast, accurate, parameter-free and user friendly 2D and 3D MFMT image formation.