Clinical Translation of Stimulated Raman Histology

受激拉曼组织学的临床转化

• 批准号: 10445765
• 负责人: Daniel Orringer
• 金额: $ 49.29万
• 依托单位: NEW YORK UNIVERSITY SCHOOL OF MEDICINE
• 依托单位国家: 美国
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-08-01 至 2027-07-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10445765
• 关键词: ATRX gene; Adjuvant; Algorithms; Artificial Intelligence; Artificial Intelligence platform; Biopsy; Brain Neoplasms; Classification; Clinical; Clinical Data; Clinical Trials; Clinical Trials Design; Complex; Computer software; Data; Development; Diagnosis; Diagnostic; Diffuse; Epigenetic Process; Excision; Genetic; Glioma; Goals; Guidelines; Histologic; Histology; Human; Image; Infrastructure; Label; Link; Malignant Neoplasms; Medicine; Methods; Microscopic; Molecular; Molecular Diagnosis; Morphology; Multi-Institutional Clinical Trial; Mutation; Natural Language Processing; Nature; Operating Rooms; Operative Surgical Procedures; Optics; Pathology; Pathology Report; Patient Care; Patients; Radiation Tolerance; Specimen; Study Subject; Supervision; System; Time; Training; Training and Infrastructure; Work; algorithm training; artificial intelligence algorithm; automated algorithm; base; central database; clinical translation; convolutional neural network; data exchange; data pipeline; deep neural network; design; diagnostic screening; digital pathology; feature extraction; graph neural network; histological image; imaging system; improved; industry partner; learning strategy; long short term memory; molecular diagnostics; molecular marker; prediction algorithm; prospective; repository; supervised learning; targeted treatment; tumor; tumor diagnosis

Molecular classification has transformed the diagnosis and treatment of brain tumors and created promising avenues for targeted therapies. However, the clinical impact of molecular classification for brain tumor patients has been blunted by long turnaround times (days-weeks), as well as the complex infrastructure and workflow required to access genetic and epigenetic data from clinical specimens. A system for rapid (<2 minute) molecular diagnostic screening would redefine the surgical and non-surgical treatment of diffuse gliomas. Rapid intraoperative molecular classification would identify the patients who would benefit most from radical resection from those with chemo- and/or radio-sensitive tumors where a more conservative surgical approach, relying more heavily on adjuvant treatment, might be best. Immediate molecular classification would also facilitate the use of targeted therapeutics and transform the way clinical trials in the glioma field are designed by enabling rapid identification of potential study subjects based on molecular criteria during or shortly after biopsy. Through our academic-industrial partnership with Invenio Imaging Inc, we developed, implemented and validated an accurate bedside system for rapid morphologic diagnosis through label-free stimulated Raman histologic (SRH) imaging paired with an artificial intelligence-based algorithm (Hollon et al. Nature Medicine 2020). Here, we intend to leverage and enhance our unique AI-based platform for intraoperative diagnosis to predict the molecular alterations that define diffuse gliomas within minutes of biopsy in the operating room without the need for a pathology lab or specialized analytic facility. This proposal is expected to yield an autonomous diagnostic workflow for human brain tumor molecular diagnostics. The work proposed here represents the development of a platform approach by which molecular diagnosis is unlocked through SRH and artificial intelligence, thus creating a new standard of accessibility for molecular diagnosis in human cancer.

分子分类改变了脑肿瘤的诊断和治疗，创造了光明的前景 靶向治疗的途径。然而，分子分类对脑肿瘤患者的临床影响 由于周转时间较长(几天到几周)以及复杂的基础架构和工作流程， 需要从临床样本中获取遗传和表观遗传学数据。 快速(2分钟)分子诊断筛查系统将重新定义外科和非外科手术 弥漫性胶质瘤的治疗。术中快速分子分类将识别哪些患者 从根治性切除中获益最大的是那些对化疗和/或放射敏感的肿瘤 保守的手术方法，更多地依赖于辅助治疗，可能是最好的。立马 分子分类还将促进靶向治疗的使用，并改变临床方式 胶质瘤领域的试验设计是通过基于以下条件快速识别潜在研究对象 活组织检查期间或术后不久的分子标准。 通过我们与Invenio Image Inc.的学术-产业合作伙伴关系，我们开发、实施和 通过无标记受激拉曼验证了用于快速形态诊断的准确床边系统 组织学(SRH)成像与基于人工智能的算法配对(Hollon等人)。自然医学 2020)。在这里，我们打算利用和增强我们独特的基于人工智能的术中诊断平台来 在活检后几分钟内预测弥漫性胶质瘤的分子变化 手术室，不需要病理实验室或专门的分析设备。 这项提议有望为人脑肿瘤分子产生一个自主的诊断工作流程 诊断。这里提出的工作代表了一种平台方法的发展，通过这种方法，分子 通过SRH和人工智能解锁诊断，从而创建可访问性的新标准 人类癌症的分子诊断。