Placental Pathology: Digital Assessment and Validation

描述（由申请人提供）：羊膜内感染是早产的主要危险因素（围产期发病率和死亡率的最大单一因素），也是脑瘫和其他重大儿童疾病发展的一个因素。因此，其诊断需要既可靠（可在同一患者中重复，也可在不同患者和机构中重复）又有效（一致地预测感染的重要临床特征，包括严重程度、持续时间和新生儿败血症等后遗症的风险）。不幸的是，目前的诊断病理学“金标准”既不可靠，也没有根据“群体共识”以外的措施进行验证，而“群体共识”是生物学上有效的终点（如羊水或脐带血蛋白质组学）的糟糕替代品。只有少数美国病理学家拥有这种诊断的专业知识，其中大多数医生位于学术医疗中心，限制了可以提供服务的婴儿的绝对数量以及可以提供这种护理的地点。因此，美国绝大多数妇产医院实际上缺乏提供此类护理所需的专家人员，导致只有一小部分新生儿能够准确可靠地评估羊膜内感染的风险。在这个第一阶段的建议中，我们将首先使我们的算法对苏木精和伊红染色的可变性具有鲁棒性，这将在不同医院实验室制备的载玻片中得到预期。接下来，为了验证这些措施，我们将利用2个大型数据集，其中胎盘已经被收集、采样、切片和染色，幻灯片已经被评分（使用当前的标准方法）和数字化，以及一个参与数据集并致力于该项目的专家病理学家团队。利用这些资源，我们将确定我们的产品的可靠性，图像分析软件是一个基于算法的图像分割工具工具箱，可以取代常规苏木精和伊红（H&E）染色载玻片中目前的“最佳实践”（中性粒细胞浸润的半定量）。针对专家病理学家和生物学相关终点（与感染/炎症相关分子相关的羊水或脐带血蛋白质组学）的验证，将为该诊断工具推向市场做好准备，为胎盘病理学家这一小群“专家”无法触及的婴儿提供最先进的诊断护理。实际上，这一工具将开启可靠、可重复和有效诊断的潜力，在世界上任何一家可以生产苏木精和伊红染色载玻片的医院进行。公共卫生相关性：胎儿炎症反应，定义为脐带血中炎性细胞因子水平升高以及胎盘脐带和绒毛膜血管炎，可预测早产（优化下次妊娠结局）的复发风险，以及更普遍的脑室内出血、脑白质损伤、脑瘫、儿童哮喘和肺损伤的风险。目前用于测量炎症的组织学工具仅限于中性粒细胞数量的半定量估计，即使在“专家”中也具有有限的可重复性。胎盘分析公司开发了一套用于数字化组织学切片的图像处理算法，有望可靠地量化当前羊膜内感染组织学评估的“金标准”（即中性粒细胞数量），以及中性粒细胞核分裂的量化，这可能比任何当前的组织学方法都更精确地与年龄/感染持续时间相关。

DESCRIPTION (provided by applicant): Intraamniotic infection is a major risk factor for preterm birth (the greatest single agent of perinatal morbidity and mortality) as well as a contributor to the development of cerebral palsy and other significant childhood diseases. As such, its diagnosis needs to be both reliable (reproducible in the same patient, and across patients and institutions) and valid (consistently predictive of important clinical features of infection, including severity, duration and risk of sequelae such as neonatal sepsis). Unfortunately, current diagnostic pathology "gold standards" are neither reliable nor have they been validated against measures other than "group consensus", a poor substitute for biologically valid endpoints such as amniotic fluid or cord blood proteomics. Only a handful of US pathologists possess expertise in this diagnosis, with most of these physicians located in academic medical centers, limiting both the absolute numbers of infants who can be provided services as well as where such care can be provided. The vast majority of US maternity hospitals therefore effectively lack access to the expert personnel needed to provide such care, resulting in a system in which only a small proportion of newborns can be accurately and reliably assessed for exposure to intraamniotic infections. In this Phase 1 proposal, we will, first, make our algorithms robust to the variability in hematoxylin and eosin staining that would be expected in slides prepared from diverse hospital laboratories. Next, in order to validate these measures, we will take advantage of 2 large data sets in which placentas have already been collected, sampled, sectioned and stained, and slides have been scored (using current standard methods) and digitized, and a team of expert pathologists who have been involved with the data sets and are committed to the project. Using these resources, we will determine the reliability of our product, image analysis software that is a toolbox of algorithm-based image segmentation tools that can replace the current "best practice" (semi-quantitation of neutrophil infiltrates) in routine hematoxylin and eosin (H&E) stained slides. Validation against both expert pathologists and biologically germane endpoints (amniotic fluid or cord blood proteomics related to infection/inflammation associated molecules) will prepare this diagnostic tool for market introduction to provide state of the art diagnostic care for infants beyond the reach of the small cadre of "expert" placental pathologists. In effect, this tool will open the potential for reliable, reproducible and valid diagnoses to be performed at any hospital in the world that can produce a hematoxylin and eosin stained slide. PUBLIC HEALTH RELEVANCE: The fetal inflammatory response, defined as elevated levels of inflammatory cytokines in cord blood and by vasculitis in the umbilical and chorionic vessels of the placenta, predicts recurrence risk for preterm birth (optimizing next pregnancy outcomes), and risks of intraventricular hemorrhage, cerebral white matter damage, cerebral palsy, childhood asthma and lung damage more generally. The current histologic tools used to measure inflammation are limited to a semiquantative estimation of neutrophil number with limited reproducibility even among "experts". Placental Analytics has developed a set of image processing algorithms for digitized histology slides that promises reliable quantification of the current "gold standard" for histologic assessment of intraamniotic infection (i.e., neutrophil number), as well as quantification of neutrophil karyorrhexis that may correlate with age/duration of infection more precisely than any current histologic method.