CONTEXT INFORMATION IN AUTOMATED CERVICAL SMEAR ANALYSIS

这项研究的长期目标是开发和测试自动化 细胞图像分析在提高宫颈病变诊断和预后中的作用 癌症。我们之前的研究表明，语境分析(LOW 场景的分辨率分析)和标记分析(形态变化 在外观正常的单元格中)将重要信息添加到HIGH 乳腺、前列腺和宫颈的分辨率个体细胞分析 肿瘤。之前的研究分几个步骤进行， 效率低下，只允许我们收集较小的数据集。 拟议研究的主要目标是开发一种 集成分析系统(IAS)将采用更复杂的 方法将单元格分析、上下文分析和标记分析相结合。这个 这项研究的第一个目标是开发一种系统，允许同时 单层的上下文分析、标记分析和单个细胞分析 宫颈脱落细胞的制备。这将包括使用 自适应方法，结合先验信息来指导 分析过程。第二个目标是测试组合方法论。 以确定它是否比任何单一的分析提供了改进 方法。第三个目标是应用这些技术来调查 区分持续性/渐进性与倒退性和渐进性的可行性 阴性病例。 这项研究应该对成本控制和 宫颈细胞学在宫颈癌预筛查中的应用 自动化细胞学可以在一定程度上弥补日益严重的合格细胞学短缺 细胞技术专家。它还可以提供更客观的标准 不同经验层次宫颈肿瘤的诊断和预后 和训练。同样的方法最终也可以应用于专家 用于诊断和教学的系统。此外，在这方面取得的进展 研究应该可以很容易地转移到其他类型的人类癌症。 因此，我们预计这项研究将对 癌症预防和控制的实用和科学方面。

The long-term objective of this research is to develop and test automated cell image analysis for the improved diagnosis and prognosis of cervical cancer. our previous research has shown that contextual analysis (low resolution analysis of scenes) and marker analysis (morphometric changes in "normal" looking cells) adds significant information to high resolution individual cell analysis of breast, prostate, and cervical neoplasia. The previous studies were performed in several steps, which was inefficient and only permitted us to collect small data sets. The primary objective of the proposed research is to develop an Integrated Analysis System (IAS) that will employ more sophisticated methods to combine the single cell, contextual, and marker analyses. The first aim of this research is to develop a system to permit simultaneous contextual, marker, and individual cell analyses of monolayer preparations of exfoliated cervical cells. This will include the use of adaptive methodologies that incorporate a priori information to guide the analysis process. The second aim is to test the combined methodolology to determine whether it provides an improvement over any single analysis method. The third aim is to apply these technique to investigate the feasibility of discriminating persistent/progressive from regressive and negative cases. This research should have significant impact on cost-containment and availability of cervical cytology for pre-screening of cervical cancer Automated cytology can partly offset the growing shortage of qualified cytotechnologists. It can also provide more objective criteria for diagnosis and prognosis of cervical neoplasia at all levels of experience and training. The same methodology can eventually be applied to expert systems for diagnosis and teaching. Moreover, advances made in this research should be readily transferable to other types of human cancer. Thus, we expect that this research will have far-reaching effects on practical as well as scientific aspects of cancer prevention and control.