Computer-Aided Analysis of Histopathology Images of Prostate Cancer

描述（由申请人提供）：该应用程序的广泛，长期目标是通过早期检测和诊断这种疾病来根除作为死亡率主要原因的前列腺癌。该项目的目标是测试开发一种新技术的可行性，该技术挑战了前列腺癌组织病理学临床实践中的现有范式，使用计算机图像分析来辅助病理学家进行前列腺癌诊断。待检验的假设是，计算机技术可以被开发来准确地分析前列腺癌组织病理学图像。具体目标是：（1）建立前列腺癌数字组织病理学图像数据库;（2）开发AMACR染色图像中前列腺癌自动识别方法;（3）开发结构和形态模式定量分析方法;（4）开发Gleason等级自动评分方法。研究设计将与泌尿病理专家合作建立数字图像数据库，开发用于前列腺组织病理学图像的计算机图像分析技术，并进行自动前列腺癌检测和分级的可行性测试，以帮助病理学家。所使用的方法包括数字图像采集、图像处理、彩色图像分析、图像分割、观察者研究和计算机性能评估。追求这些目标的理由包括先进的计算机辅助诊断的医学影像解释的放射科医生，在开发计算机辅助诊断方法的研究团队的经验，从模拟成像到数字成像在临床组织病理学的迁移。这项研究与公共卫生有关，因为虽然作为其他医学分支的黄金标准，但组织病理学对前列腺癌的诊断并不是百分之百准确。组织病理学图像的计算机辅助分析可能使前列腺癌的组织病理学诊断更准确，这可能导致更有效的癌症治疗和更好的无癌生存结局。如果该应用的目标得以实现，并且如果成功证明了可行性，则将开辟一个新的、以前未探索的研究领域，并对临床实践产生潜在的未来影响。

DESCRIPTION (provided by applicant): The application's broad, long-term objective is to eradicate prostate cancer as a major cause of mortality through the early detection and diagnosis of this disease. The goal of this project is to test the feasibility of developing a new technology that challenges the existing paradigm in the clinical practice of prostate cancer histopathology, to use computer image analysis for assisting pathologists in prostate cancer diagnosis. The hypothesis to be test is that computer techniques can be developed to analyze prostate cancer histopathology images accurately. The specific aims are: (1) to establish a database of digital histopathology images of prostate cancer; (2) to develop methods for automated identification of prostate cancer in AMACR-stained images; (3) to develop methods for quantitative analysis of architectural and morphological patterns; and (4) to develop methods for automated scoring of Gleason grades. The research design will be to build a digital image database in collaboration with an expert urologic pathologist, develop computer image-analysis techniques for prostate histopathology images, and to conduct feasibility tests for automated prostate cancer detection and grading as an aid to pathologists. The methods to be used include digital image acquisition, image processing, color image analysis, image segmentation, observer study, and computer performance evaluations. The rationales for pursuing these goals include advance in computer-aided diagnosis for medical images interpreted by radiologists, the experience of the research team in developing computer-aided diagnosis methods, and a migration from analog imaging to digital imaging in clinical histopathology. This research is relevant to public health in that, while serving as a gold standard for other branches of medicine, histopathology is not one hundred percent accurate for prostate cancer diagnosis. Computer-aided analysis of histopathology images can potentially make histopathology diagnosis of prostate cancer more accurate, which could lead to more effective cancer treatment and better outcomes of cancer-free survival. If the aims of this application are achieved, and if feasibilities are demonstrated successfully, then a new, previously not explored area of research will open up, with potential future impact to clinical practice.