Deep learning for automatic microscopy image analysis

<p>Microscopy imaging techniques allow for the creation of detailed images of cells (or nuclei) and have been widely employed for cell studies in biological research and disease diagnosis in clinic practices.Microscopy image analysis (MIA), with tasks of cell detection, cell classification, and cell counting, etc., can assist with the quantitative analysis of cells and provide useful information for a cellular-level understanding of biological activities and pathology. Manual MIA is tedious, time-consuming, prone to subject errors, and are not feasible for the high-throughput cell analysis process. Thus, automatic MIA methods can facilitate all kinds of biological studies and clinical tasks. Conventional feature engineering-based methods use handcrafted features to address MIA problems, but their performances are generally limited since the handcrafted features can lack feature diversity as well as relevancy to specific tasks. In recent years, deep learning, especially convolutional neuronal networks (CNNs), have shown promising performances on MIA tasks, due to their strong ability to automatically learn task-specific features directly from images in an end-to-end learning manner. However, there still remains a large gap between deep learning algorithms shown to be successful on retrospective datasets and those translated to clinical and biological practice. The major challenges for the application of deep learning into practical MIA problems include: (1) MIA tasks themselves are challenging due to limited image quality, the ambiguous appearance of inter-class nuclei, occluded cells, low cell specificity, and imaging artifacts; (2) training a learning algorithm is very challenging due to the potential gradient vanishing issue and the limited availability of annotated images. In this thesis, we investigate and propose deep learning methods for three challenging MIA tasks: cell counting, multi-class nuclei segmentation, and 3D phase-to-fluorescent image translation. We demonstrate the effectiveness of the proposed methods by intensively evaluating them on practical MIA problems. The proposed methods show superior performances compared to competitive state-of-the-art methods. Experimental results demonstrated that the proposed methods hold great promise to be applied in practical biomedical applications.</p>

Name thesis:degree_name

Doctor of Philosophy (PhD)

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science & Engineering

Year dc:date.available

2021

Author dc:creator

• He, Shenghua

Contributors dc:contributor

• Tao Ju
• Mark Anastasio

• Biomedical image analysis
• Computer vision
• Deep learning
• Machine learning
• Engineering

Statement dc:rights

• I have not registered my thesis with the U.S. Copyright Office, and do not intend to.

Language dc:language

English (en)

Identifier

https://openscholarship.wustl.edu/eng_etds/725

OAI identifier oai:identifier

oai:openscholarship.wustl.edu:eng_etds-1777