Metasurface enhanced and machine learning aided spectrochemical liquid biopsy

超表面增强和机器学习辅助光谱化学液体活检

• 批准号: 10647397
• 负责人: Filiz Yesilkoy
• 金额: $ 22.33万
• 依托单位: UNIVERSITY OF WISCONSIN-MADISON
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-03-01 至 2026-02-28
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10647397
• 关键词: Address; Benign; Biological Markers; Biophotonics; Blood Tests; Cancer Biology; Cancer Control; Cancer Detection; Cancer Diagnostics; Cancer Patient; Chemicals; Clinical; Clinical Management; Complex; Coupled; Data; Detection; Devices; Diagnosis; Diagnostic; Discrimination; Discrimination Learning; Disease; Early Diagnosis; Electromagnetics; Engineering; Equity; Female; Fingerprint; Functional disorder; Funding; Future; Goals; Grant; Health; Image; Imaging technology; Label; Lasers; Light; Lipids; Machine Learning; Malignant Neoplasms; Malignant neoplasm of ovary; Measurement; Medical; Methods; Modality; Modeling; Molecular; Non-Invasive Cancer Detection; Non-Invasive Detection; Nucleic Acids; Optics; Outcome; Pathologic; Patient Monitoring; Patients; Pattern; Pelvis; Performance; Peritoneal Fluid; Physiological; Population; Population Group; Population Heterogeneity; Postmenopause; Proteins; Protocols documentation; Raman Spectrum Analysis; Recommendation; Research; Research Personnel; Retrieval; Sampling; Screening for cancer; Serum; Signal Transduction; Socioeconomic Factors; Spectrometry; Spectrum Analysis; Survival Rate; Symptoms; System; Techniques; Technology; Temperature; Testing; Tissues; United States National Institutes of Health; Validation; absorption; advanced analytics; burden of illness; cancer health disparity; cancer risk; cancer type; chemical fingerprinting; cohort; cost; cost effective; data analysis pipeline; design; detection platform; diagnostic biomarker; imaging approach; improved; infrared spectroscopy; innovation; instrumentation; liquid biopsy; machine learning method; machine learning model; manufacture; mid infrared spectrometry; mortality; multiple omics; nano; nanophotonic; novel; novel diagnostics; photonics; plasmonics; portability; preclinical study; quantum; real world application; screening; screening program; spectrograph; survival outcome; tool

PROJECT SUMMARY Liquid biopsy modalities that can non-invasively detect disease-associated biomarkers from biofluids can enable early cancer detection and patient monitoring with implications for improved survival rates. However, current methods have not achieved critical sensitivity and accuracy to be approved for population screening programs. New spectrochemical liquid biopsy methods, such as Raman and infrared spectroscopy, coupled with machine learning models are emerging as next-generation diagnostic modalities. Yet, fundamental physical limitations of light-matter interactions using conventional optical setups hinder the analytical performance of molecular spectroscopy techniques. Here, we propose to employ novel electromagnetic metasurfaces that can advance the analytical sensitivity and chemical selectivity of infrared absorption spectroscopy enabling its real-world applications in the biomedical field. Moreover, our innovative laser-based spectral imaging approach can achieve on-chip spectrometer-less chemical fingerprint retrieval eliminating clinically incompatible, complex, and bulky instrumentation requirements. The long-term goal of this project is to develop a rapid, label-free, portable, and non-invasive cancer detection platform based on sensitive and accurate chemometric liquid biopsy and machine learning-aided discrimination modalities. The overall objectives in this application are to (i) determine a potent metasurface design that can robustly extract chemical fingerprint information from a complex biosample matrix, (ii) identify optimized design parameters for spectral imaging-based on-chip fingerprint retrieval (iii) establish measurement protocols and data processing pipeline (iv) identify a machine learning model by which sensitive and accurate sample discrimination can be achieved. In the short term, we will pursue two specific aims: 1) develop novel engineered metasurfaces for sensitive and specific spectrochemical biofluid analysis and demonstrate spectrometer-less on-chip chemical fingerprinting 2) Test and validate the platform using biofluids from an ovarian cancer patient cohort and non-cancer controls. Our proposed approach is innovative because it catalyzes the state-of-the-art laser-based infrared spectral imaging technology with powerful nanophotonic tools to enable its impact in biomedical diagnostics and address an unmet medical need. In addition, the proposed interdisciplinary project is significant because it is expected to develop a non-invasive and accessible health screening platform that can ultimately impact the clinical management of cancer and the survival outcomes equitably among diverse populations.

项目摘要 可以从生物流体中非侵入性地检测疾病相关生物标志物的液体活检模式可以使 早期癌症检测和患者监测，对提高生存率有意义。但目前的 这些方法还没有达到临界灵敏度和准确度，不能被批准用于人群筛查项目。 新的光谱化学液体活检方法，如拉曼和红外光谱，结合机器 学习模型正在作为下一代诊断模式出现。然而，基本的物理限制 使用常规光学装置的光-物质相互作用阻碍了分子的分析性能， 光谱技术。在这里，我们建议采用新的电磁超表面， 红外吸收光谱的分析灵敏度和化学选择性使其能够在现实世界中 在生物医学领域的应用。此外，我们创新的基于激光的光谱成像方法可以实现 片上无光谱仪化学指纹检索，消除了临床上不兼容、复杂和庞大的问题 仪表要求。 这个项目的长期目标是开发一种快速、无标签、便携、无创的癌症检测 基于灵敏准确的化学计量液体活检和机器学习辅助鉴别的平台 方式。本申请的总体目标是（i）确定一种有效的元表面设计， 从复杂生物样品基质中稳健地提取化学指纹信息，（ii）识别优化设计 用于基于光谱成像的芯片上指纹检索的参数（iii）建立测量协议， 数据处理流水线（iv）识别机器学习模型，通过该模型敏感且准确地采样 可以实现歧视。在短期内，我们将追求两个具体目标：1）开发新的工程 用于灵敏度和特异性光谱化学生物流体分析的超颖表面， 2）使用来自卵巢癌患者的生物流体测试和验证平台 队列和非癌症对照。我们提出的方法是创新的，因为它催化了最先进的 基于激光的红外光谱成像技术与强大的纳米光子工具，使其在 生物医学诊断和解决未满足的医疗需求。此外，拟议的跨学科项目 意义重大，因为它有望开发一个非侵入性和可访问的健康筛查平台， 最终公平地影响癌症的临床管理和不同人群的生存结果 人口。