【信息技术】【2007.08】图像分析技术在牛肺部疾病分类中的应用

本文为加拿大萨斯喀彻温大学（作者：C. Denise Miller）的硕士论文，共161页。

组织样本的组织学分析常常是诊断疾病的关键步骤。然而，这种类型的评估本质上是主观的，因此不同病理学家的判断结果可能出现高度的差异。组织学分析对于病理学家来说也是一项非常耗时的任务。基于计算机的组织样本定量分析能够降低传统组织评估的主观性，并潜在地减少分析每个样本所需的时间。

本研究旨在探讨影像处理技术，并开发出可作为牛肺组织病理学评估的诊断辅助软件。该软件通过检查组织样本的数字图像，识别并突出显示一系列特征，利用这些特征诊断疾病，并且可以用于区分各种肺部疾病。该软件的检查输出是一系列强调相关疾病风险的分割图像，以及量化这些特征在组织样本中进行的数据测量。对50个牛肺组织样本的软件分析结果与病理专家对这些样本的详细分析结果进行了比较。

诊断软件中实现的图像分析技术显示出实用的潜力。每种疾病的检测在某种程度上是成功的，在某些情况下的分析结果是极好的。然而，用于识别一组疾病指标的准确率具有很大差异，其灵敏度值在最高94.8%到最低22.6%之间变化。这种结果变化的部分原因是由于确定精度所用方法的局限性。

Histologic analysis of tissue samples is often a critical step in thediagnosis of disease. However, this type of assessment is inherentlysubjective, and consequently a high degree of variability may occur betweenresults produced by different pathologists. Histologic analysis is also a verytime-consuming task for pathologists. Computer-based quantitative analysis oftissue samples shows promise for both reducing the subjectivity of traditionalmanual tissue assessments, as well as potentially reducing the time required toanalyze each sample. The objective of this thesis project was to investigateimage processing techniques and to develop software which could be used as adiagnostic aid in pathology assessments of cattle lung tissue samples. Thesoftware examines digital images of tissue samples, identifying and highlightingthe presence of a set of features that indicate disease, and that can be usedto distinguish various pulmonary diseases from one another. The output of thesoftware is a series of segmented images with relevant disease indicatorshighlighted, and measurements quantifying the occurrence of these featureswithin the tissue samples. Results of the software analysis of a set of 50cattle lung tissue samples were compared to the detailed manual analysis ofthese samples by a pathology expert. The combination of image analysistechniques implemented in the thesis software shows potential. Detection ofeach of the disease indicators is successful to some extent, and in some casesthe analysis results are extremely good. There is a large difference inaccuracy rates for identification of the set of disease indicators, however,with sensitivity values ranging from a high of 94.8% to a low of 22.6%. Thiswide variation in result scores is partially due to limitations of themethodology used to determine accuracy.

1 引言

2 研究目标

3 项目背景

4 诊断数据与方法

5 研究结果

6 结论、讨论与未来工作

7 参考文献

附录A 补充材料信息

下载英文原文地址：

http://page2.dfpan.com/fs/clc9j2f21129e16a8f7/

更多精彩文章请关注微信号：