Clinical Translation of Stimulated Raman Histology

准确的术中组织诊断是脑肿瘤手术中医疗决策的核心。现有 术中组织学技术消耗了少量的组织活检，引入了冷冻伪影，并高度依赖于 在专门的外科病理学实验室工作的熟练技术人员和神经病理学家， 解释幻灯片。此外，进行脑肿瘤手术的中心数量远远超过 委员会认证的神经病理学家，消除了在许多情况下专家咨询的可能性。即使在 最先进的，人员配备齐全的医院，术中病理的周转时间构成了主要障碍， 手术护理的有效提供，突出了对快速诊断的改进系统的需要。 受激拉曼组织学（SRH）创建未处理组织的高分辨率数字显微图像 标本的一小部分时间的传统技术，消除了依赖于冷冻切片 用于切片、染色、封片和审查载玻片的实验室。虽然SRH已被证明可以揭示关键的 脑肿瘤标本的诊断组织学特征，与激光安全性相关的主要技术障碍， 性能阻碍了其临床转化。建立了现有的学术-工业伙伴关系 合作者已经导致了临床兼容的SRH显微镜的开发和初步验证， 患者护理环境（Nature Biomedical Engineering 1：0027，2017）。我们已经证明，SRH具有诊断性 与传统组织学技术相当的价值，SRH图像非常适合通过 一种自动机器学习算法 拟议的研究计划代表了多学科（神经外科，神经病理学，生物统计学， 计算机科学和医疗器械制造）的学术和产业合作，以加速发展 脑肿瘤手术中使用的SRH成像仪。伙伴关系的总体目标是为以下方面创造途径： 外科医生和病理学家之间使用SRH成像仪的在线合作，为 SRH在前瞻性随机对照试验中的效用，并提高我们的能力， 采用最先进的图像分类方法的自动术中诊断， 神经网络 一旦完成，这项研究计划将通过简化流程来改善脑肿瘤患者的护理 用于术中诊断。它还将为远程和自动诊断创造一条途径， 在神经病理学方面的研究成果，用于更多的治疗脑瘤的中心。SRH技术结合算法， 自动化术中诊断为推进更大的外科肿瘤学领域创造了可能性， 组织学信息对于做出手术决定是必不可少的。

Accurate intraoperative tissue diagnosis is central to medical decision making during brain tumor surgery. Existing intraoperative histologic techniques deplete scant tissue biopsies, introduce freezing artifacts, and rely on highly- skilled technicians and neuropathologists working in dedicated surgical pathology laboratories to produce and interpret slides. In addition, the number of centers where brain tumor surgery is performed far exceeds the number of board-certified neuropathologists, eliminating the possibility for expert consultation in many cases. Even in the most advanced, well-staffed hospitals, turnaround time for intraoperative pathology poses a major barrier for the efficient delivery of surgical care, highlighting the need for an improved system for rapid diagnosis. Stimulated Raman histology (SRH) creates high-resolution digital microscopic images of unprocessed tissue specimens in a fraction of the time of conventional techniques and eliminates reliance on a frozen section laboratory for sectioning, staining, mounting and reviewing slides. While SRH has been shown to reveal key diagnostic histologic features in brain tumor specimens, major technical hurdles related to laser safety and performance have hindered its clinical translation. The existing academic-industrial partnership between established collaborators has resulted in the development and initial validation of a clinically-compatible SRH microscope in a patient care setting (Nature Biomedical Engineering 1:0027, 2017). We have demonstrated that SRH has diagnostic value comparable to conventional histologic techniques and that SRH images are well-suited for interpretation via an automated machine learning algorithm. The proposed research program represents a multi-disciplinary (neurosurgery, neuropathology, biostatistics, computer science and medical device manufacturing) academic-industry partnership to accelerate the development of an SRH imager for use during brain tumor surgery. The overall goal of the partnership is to create pathways for online collaboration between surgeons and pathologists using the SRH imager, provide strong evidence for the utility of SRH in the setting of a prospective randomized controlled trial and to enhance our capabilities for automated intraoperative diagnosis employing state-of the-art image classification methods including convolutional neural networks. Once completed, this research program will improve the care of brain tumor patients by streamlining the process for intraoperative diagnosis. It will also create a pathway for remote and automated diagnosis, extending expertise in neuropathology to more centers caring for brain tumor patients. SRH technology combined with algorithms for automated intraoperative diagnosis create the possibility of advancing the larger field of surgical oncology where histologic information is essential for making surgical decisions.