ERI: A New Spatial Image Analysis of the Tumor Microenvironment for Cancer Immunotherapy

癌症免疫疗法利用人体免疫系统的力量对抗癌细胞。最近的进展揭示了癌症的起始和反应，关键疗法在癌症患者中显示出显着的有效性。然而，对癌症免疫治疗的反应高度依赖于癌细胞与其微环境之间的相互作用。该工程研究启动（ERI）奖将解决一个重要的挑战——利用基于人工智能的方法探索、分析和解释肿瘤微环境中的复杂相互作用。作为该项目的一部分，K-12，本科生和研究生将通过协同研究和教育活动介绍生物医学工程概念。该项目的长期研究目标是发现具有统计学意义的癌症生物标志物，以帮助研究人员确定癌症患者的预后。为了实现这一目标，本项目旨在通过双通道管道设计来揭示异质性癌症相关群体之间的定量空间关系，以实现两个目标：(i)利用从整个幻灯片图像中提取的深度学习特征来表征不同的细胞类型，如肿瘤、肿瘤浸润淋巴细胞和基质；（ii）研究具有统计学意义的免疫细胞的异质性癌症相关群体之间的空间相关性。与传统的组织病理学空间图像分析方法将图像区域划分为两个或多个类别不同，该方法覆盖了来自异质癌症相关群体的分类结果，以主动确定不同癌症相关群体细胞在高水平浸润免疫细胞和低水平浸润免疫细胞中的空间相关性。这项研究将广泛地促进基于人工智能的生物标志物开发、多种类型癌症的研究以及其他研究领域的空间变异解释。这个项目的结果有望产生新的知识，以帮助预测对癌症免疫治疗的反应。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Cancer immunotherapy uses the power of the body’s immune system against cancer cells. Recent progress has shed light on cancer initiation and response, with key therapies showing significant effectiveness in cancer patients. However, response to cancer immunotherapy is highly dependent on interaction between cancer cells and their microenvironment. This Engineering Research Initiation (ERI) award will tackle an important challenge — exploring, analyzing, and interpreting the complex interactions in the tumor microenvironment with AI-based approaches. As part of this project, K-12, undergraduate, and graduate students will be introduced to biomedical engineering concepts through synergistic research and educational activities. The long-term research goal of this project is to discover statistically significant cancer biomarkers that will aid researchers in determining the prognosis of cancer patients. Towards this goal, this project aims to uncover the quantitative spatial relationship between heterogeneous cancer-associated populations using a two-pass pipeline design to achieve two objectives: (i) to characterize different cell types such as tumor, tumor-infiltrating lymphocytes, and stroma with deep-learned features extracted from whole slide images and (ii) to investigate spatial correlations between heterogeneous cancer-associated populations with statistically significant immune cells. Unlike conventional histopathological spatial image analysis methods that classify image regions into two or more categories, the proposed method overlays the results of the classification from heterogeneous cancer-associated populations to proactively determine the spatial correlations of different cancer-associated populations cells in both the high level of infiltrating immune cells and the low level of infiltrating immune cells. The proposed research will broadly benefit AI-based biomarker development, investigation of multiple types of cancer, and interpretation of spatial variation in other research fields. The outcomes of this project are expected to yield new knowledge to help predict response to cancer immunotherapy.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.