Automated quantification of lipid droplets in fatty liver tissue sections

描述（由申请人提供）：非酒精性脂肪肝与肥胖、HCV感染相关，是纤维化、肝硬化和肝癌的主要原因。脂滴形成（脂肪变性）是病理的潜在原因。然而，病理学家对脂肪变性和纤维化的评分存在相当大的差异。本提案为STTR项目1R41DK082087-01“脂肪肝组织切片脂滴自动定量”的二期项目。在第一阶段，我们开发了一种图像分析算法（脂肪变性算法），通过分析苏木精+伊红染色的人体活检获得的数字照片，来预测与该项目相关的病理学家的脂肪变性评分。对于II期，Vala Sciences Inc .与知名病理学家和肝脏专家团队合作，进一步开发脂肪变性算法，并开发量化炎症和纤维化的算法。为此，我们建议使用自动数字幻灯片扫描仪扫描大量的幻灯片，以提供数千张数字图像用于训练算法。扫描的载玻片组包括：1)来自HALT-C临床试验的载玻片，其中聚乙二醇化干扰素在一项纵向研究中对HCV的可能治疗效果进行了测试；2)来自与该项目相关的病理学家和医生先前收集和保存的酒精性和非酒精性脂肪肝和脂肪性肝炎患者的载玻片；3)肝脏代谢研究中获得的小鼠肝脏载玻片。该算法将为从人类和动物脂肪肝模型中获得的肝脏样本提供脂肪变性、炎症和纤维化的客观评估，并将用于临床和研究领域。

DESCRIPTION (provided by applicant): NonAlcoholic Fatty liver is associated with obesity, and HCV infection, and is a leading cause of fibrosis, cirrhosis, and liver cancer. Lipid droplet formation (steatosis) is an underlying cause of the pathologies. However, there is considerable variation in the scoring of steatosis and fibrosis by pathologists. This proposal is for Phase II of the STTR project 1R41DK082087-01 "Automated quantification of lipid droplets in fatty liver tissue sections". In Phase I, an image analysis algorithm (Steatosis Algorithm), was developed to predict the Steatosis Score of the pathologist associated with the project, via the analysis of digital photographs obtained from hematoxylin + eosin stained human biopsies. For phase II, a collaboration is proposed between Vala Sciences Inc and a team of eminent pathologists and liver specialists to further develop the Steatosis Algorithm, and to develop algorithms to quantify inflammation, and fibrosis. For this purpose we propose to scan extensive sets of slides with an automated digital slide scanner, to provide thousands of digital images for use in training the algorithms. Slide sets to be scanned include: 1) slides from the HALT-C clinical trial in which pegylated interferon was tested for possible therapeutic effects against HCV in a longitudinal study, 2) slides from patients with both Alcoholic and NonAlcoholic fatty liver and Steatohepatitis previously collected and maintained by the pathologists and physicians associated with this project, and 3) mouse liver slides obtained in studies of liver metabolism. The algorithms will provide an objective assessment of steatosis, inflammation, and fibrosis, for liver samples obtained from both humans and animal models of fatty liver, and will be used in both the clinical and research areas. PUBLIC HEALTH RELEVANCE: "Fatty liver disease" is a condition that affects a high proportion of the US population, particularly people that are overweight or obese, or are infected with Hepatitis C virus. Fatty liver is characterized by the occurrence of fat droplets within the cells that make up the liver. Tissue samples (biopsies) are commonly taken from livers as part of the protocol for research studies on fatty liver disease, and the fat content, degree of inflammation, and fibrous nature of the samples are graded by pathologists, using microscopes and grading criteria, that is, at best, semi-quantitative. The proposed research will develop a technique for rapid and precise automatic quantification of the fat content of liver biopsies, which will be an aid to pathologists for diagnosing fatty liver, and to researchers investigating therapeutics to fatty liver in animal models of the disease.