基于深度学习的线粒体受药和病态细胞识别

发布时间：2024-05-12 14:26

　　据观察线粒体疾病一般是由线粒体DNA或天然DNA的遗传或突变引起的,这些疾病会使线粒体中的蛋白质或RNA分子的原始功能受到影响。线粒体细胞疾病有可能干扰生物体的正常功能,甚至导致生物体死亡,因此,有必要对线粒体细胞疾病进行检测,找出预防措施。通过线粒体细胞显微图像可以检测出其细胞形态,综合分析这些图像可以用来检测患病细胞,进而对线粒体疾病的未来行为进行预测与分类。面对这一问题人眼难以准确分辨这些细胞图像中的细微差别,而人工智能可以在检测这些图像中隐藏的固有模式方面发挥重要作用,通过对显微镜图像数据智能化分析,可以提前识别疾病,从而为临床和疾病问题提供解决方案。鉴于正常和受影响的线粒体细胞具有不同的形态特征,疾病改变了线粒体细胞的形态,检测细胞形态的变化以及与此变化相关的时间具有重要的生物学意义。我们主要研究分析线粒体细胞的显微图像,主要研究成果和创新点如下:1提出了一种正常的和药物处理的细胞图像关联分析算法,简称为IC。实验结果表明,该方法具有较好的相关性。2.提出了一种基于卷积神经网络的正常和药物处理的细胞图像识别算法·通过实验验证了分类的准确性,并与传统的方法进行了比较。3.分析了...

【文章页数】：128 页

【学位级别】：博士

【文章目录】：
摘要
Abstract
ACKNOWLEDGEMENTS
1 INTRODUCTION
    1.1 RESEARCH PROBLEM
    1.2 RESEARCH MOTIVATION
    1.3 AIMS AND OBJECTIVES
        1.3.1 Aim
        1.3.2 Hypothesis
        1.3.3 Objective
    1.4 METHODOLGY
    1.5 THESIS CONTRIBUTIONS
    1.6 THESIS ORGANIZATION
2 MITOCHONDRIAL CELL
    2.1 INTRODUTCION
    2.2 MITOCHONDRIAL CELL
    2.3 MITOCHONDRIAL CELL FUNCTION
    2.4 MITOCHONDRIAL CELL PARTS
        2.4.1 Mitochondrial DNA
        2.4.2 Mitochondrial Membranes
        2.4.3 Mitochondrial Spaces
        2.4.4 Mitochondrial Reproduction
        2.4.5 Ribosomes
        2.4.6 Cell Nucleus
    2.5 TWO PHOTON EXCITED FLUORESCENCE IMAGES
    2.6 MITOCHONDRIAL CELLS IMAGE CLASSIFICATION TECHIQUES
    2.7 DRUG AND NORMAL CELL
    2.8 DISEASES AND NORMAL CELL
    2.9 MITOCHONDRIAL MOVEMENT
    2.10 CONCLUSION
3 NORMAL,DRUG TREATED CELL IMAGES RECOGNATION ANDCORRELATION
    3.1 INTRODUCTION
    3.2 CELL RECOGNITION
        3.2.1 Data Set
        3.2.2 Cell Recognition Method
        3.2.3 Results of Cell Recognition Method
    3.3 IMPROVE DIGITAL IMAGE CORRELATION
        3.3.1 Data Set
        3.3.2 Method
        3.3.3 Result of Improve Digital Image Correlation (IC)
    3.4 CONCLUSION
4 DRUG AND NORMAL CELLS IMAGES CLASSIFICATION (DNCIC)
    4.1 INTRODUCTION
    4.2 METHOD
        4.2.1 Data Set
        4.2.2 Method Explanation
    4.3 RESULTS
        4.3.1 Newly Developed CNN Method for Mitochondrial Cell Images with High Accuracy
        4.3.2 Classification of Drug Treated and Normal cells
    4.4 DISCUSSION
    4.5 CONCLUSION
5 NORMAL AND DISEASED CELLS CLASSIFICATION (NDCC)
    5.1 INTRODUCTION
    5.2 METHOD
        5.2.1 Data Set
        5.2.2 Normal and Diseases Cells Classification
        5.2.3 CNN Algorithm
    5.3 RESULTS
        5.3.1 Heterogeneity Between Cells
        5.3.2 Variation Between Normal and Diseases Cells
        5.3.3 Histogram of Features and Classification of Diseased and Normal Cells
        5.3.4 Distinguishes Diseases Patches
        5.3.5 Model Development Diseases Patches
        5.3.6 Classification of Diseases and Normal Cell Region
    5.4 DISCUSSION
    5.5 CONCLUSION
6 MITOCHONDRIAL ORGANELLE MOVEMENT CLASSIFICATION(MOMC)
    6.1 INTRODUCTION
    6.2 METHOD
        6.2.1 Mitochondria Organelle Movement Classification (MOMC)
        6.2.2 Description of CNN Inception-V3 for Classification of Different Shape of Mitochondria (fission and fusion)
        6.2.3 Proposed Method
        6.2.4 Statistics and Reproducibility
    6.3 RESULTS
        6.3.1 Data Set
        6.3.2 Mitochondrial Dynamics
        6.3.3 Classification of Mitochondria Shape Whole Slide Images
    6.4 DISCUSSION
    6.5 CONCLUSION
7 CONCLUSIONS
    7.1 KEY CONCLUSIONS OF THE RESEARCH
    7.2 FUTURE SCOPE
REFERENCES
PUBLICATIONS



本文编号：3971450