融合案例推理与分类器的乳腺肿瘤辅助诊断
发布时间:2021-10-13 23:09
本研究采用朴素贝叶斯,K最近邻(K-NN)和基于案例推理(CBR)的方法构建了乳腺癌(BC)诊断模型,目的是通过提高准确率来优化CBR的检索过程。根据医疗专家的建议和帮助,我们选择了2014年至2016年莫桑比克马普托市中心医院(HCM)莫桑比克乳腺癌数据集的样本数据。在乳腺癌诊断数据库中选择了约1200名患者作为病例,从中产生了培训和测试集。因此,本文研究了一种将朴素贝叶斯(Na?ve Bayes),CBR和KNN相结合的乳腺癌诊断智能模型。实施该模式的主要步骤包括:(1)采用朴素贝叶斯模型将集合分为两类(2)将K-NN算法应用于CBR中以检索大多数类似情况。在第一阶段,朴素贝叶斯被用来估计患者是否有恶性肿瘤或良性肿瘤,并与K-NN和J48决策树分类器进行比较,朴素贝叶斯表现出优异的表现,准确率为95%。在第二阶段我们测试了选定的k值,结果显示99%的准确度。实施K-NN后提出的诊断框架的检索结果显示,检索到的病例之间的相似性比率高,最小距离一直低至0.13。结果显示实施的模型能够整合朴素贝叶斯和CBR用于乳腺癌诊断。它可以为卫生从业人员提供乳腺癌诊断的支持系统,从而减少诊断不准确性...
【文章来源】:合肥工业大学安徽省 211工程院校 教育部直属院校
【文章页数】:79 页
【学位级别】:硕士
【文章目录】:
Acknowledgements
ABSTRACT
摘要
Chapter 1 Introduction
1.1 Background
1.2 Problem Statement
1.3 Situation in Mozambique and China
1.4 Research Structure
Chapter 2 Literature Review
2.1 Breast Cancer Overview
2.2 Case-Based Reasoning (CBR) in healthcare
2.2.1 CBR in Breast Cancer
2.2.2 Case retrieval methods in CBR
Chapter 3 Proposed Conceptual Design and Model Implementation
3.1 Data Exploration
3.2 Descriptive Model (Phase I)
3.2.1 Na?ve Bayes
3.2.2 J48 Decision trees Classifier
3.2.3 K-NN Classifier
3.3 K-NN based CBR Retrieval (Phase II)
3.3.1 Nearest Neighbor Classifiers
3.3.2 Indices/ Indexing
3.3.3 Case retrieval
3.3.4 Case Adoption
Chapter 4 Results and Discussion
4.1 Data Preprocessing
4.1.1 Dimensionality reduction / Feature subset selection
4.1.2 Discretization and Binarization
4.1.3 Variable transformation
4.1.4 Feature Creation
4.2 Phase I
4.2.1 Evaluation
4.2.2 Comparative Performance Analysis of the Classifiers
4.3 Phase II
4.3.1 Evaluation performance of K-NN
4.3.2 CBR retrieval results
Chapter 5 Conclusion
5.1 Overall Summary
5.2 Limitations
5.3 Future Research
References
List of Academic Activities and Achievements during the Degree
【参考文献】:
期刊论文
[1]案例推理的故障诊断技术研究综述[J]. 柳玉,贲可荣. 计算机科学与探索. 2011(10)
[2]CBR技术在Multi-Agent故障诊断中的应用[J]. 朱群雄,刘光. 计算机工程与应用. 2004(21)
[3]基于范例推理的结核病专家系统[J]. 张治洪,童溶,王仲元,王巍. 天津理工学院学报. 1997(03)
本文编号:3435581
【文章来源】:合肥工业大学安徽省 211工程院校 教育部直属院校
【文章页数】:79 页
【学位级别】:硕士
【文章目录】:
Acknowledgements
ABSTRACT
摘要
Chapter 1 Introduction
1.1 Background
1.2 Problem Statement
1.3 Situation in Mozambique and China
1.4 Research Structure
Chapter 2 Literature Review
2.1 Breast Cancer Overview
2.2 Case-Based Reasoning (CBR) in healthcare
2.2.1 CBR in Breast Cancer
2.2.2 Case retrieval methods in CBR
Chapter 3 Proposed Conceptual Design and Model Implementation
3.1 Data Exploration
3.2 Descriptive Model (Phase I)
3.2.1 Na?ve Bayes
3.2.2 J48 Decision trees Classifier
3.2.3 K-NN Classifier
3.3 K-NN based CBR Retrieval (Phase II)
3.3.1 Nearest Neighbor Classifiers
3.3.2 Indices/ Indexing
3.3.3 Case retrieval
3.3.4 Case Adoption
Chapter 4 Results and Discussion
4.1 Data Preprocessing
4.1.1 Dimensionality reduction / Feature subset selection
4.1.2 Discretization and Binarization
4.1.3 Variable transformation
4.1.4 Feature Creation
4.2 Phase I
4.2.1 Evaluation
4.2.2 Comparative Performance Analysis of the Classifiers
4.3 Phase II
4.3.1 Evaluation performance of K-NN
4.3.2 CBR retrieval results
Chapter 5 Conclusion
5.1 Overall Summary
5.2 Limitations
5.3 Future Research
References
List of Academic Activities and Achievements during the Degree
【参考文献】:
期刊论文
[1]案例推理的故障诊断技术研究综述[J]. 柳玉,贲可荣. 计算机科学与探索. 2011(10)
[2]CBR技术在Multi-Agent故障诊断中的应用[J]. 朱群雄,刘光. 计算机工程与应用. 2004(21)
[3]基于范例推理的结核病专家系统[J]. 张治洪,童溶,王仲元,王巍. 天津理工学院学报. 1997(03)
本文编号:3435581
本文链接:https://www.wllwen.com/yixuelunwen/zlx/3435581.html