- Title
- Analysis of microscopy images: modelling and reconstruction
- Creator
- Abdolhoseini, Mahmoud
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2019
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- Computer-aided image analysis plays a vital role in modern medicine by introducing algorithms and techniques to extract the desired information and features out of raw images. This valuable information helps clinicians and researchers to get a better insight into disorders in different organs, such as brain tumour, lung and breast cancer, and treat them more successfully. In neuroscience, the analysis of cells and vessels in the brain highly depends on the microscopy image analysis. This thesis contributes to this field in three parts: 1- modelling of microscopy images and generating synthetic image dataset, 2- proposing a novel algorithm to segment clump of nuclei from histopathological images, 3- proposing a new tracing algorithm to reconstruct and quantify microglial cells. In the first part of this thesis we focus on modelling microscopy images and generating realistic synthetic images with known ground truths which offer a means of assessing algorithm performance. We propose a synthesizer for neuron nucleus micro-environment using Gaussian mixture model (GMM) and Perlin noise function. Nucleus shapes are generated by spline interpolation of random points on elliptical shapes. The textures of the foreground (nuclei) and the background are generated via Perlin noise, and then assigned intensities generated by applying GMM to the real data. The cell orientations are also implemented via Perlin noise to mimic the behaviour of the actual neuron cells. In the second part of this thesis we propose a novel method of segmentation applicable to variety of histopathological images stained for different proteins, with high speed, accuracy and level of automation. Our algorithm is initiated by applying a new locally adaptive thresholding method on watershed regions. Followed by a new splitting technique based on multilevel thresholding and the watershed algorithm to separate clustered nuclei. Finalized by a model-based merging step to eliminate oversegmentation and a model-based correction step to improve segmentation results. In the last part of this thesis we propose an automated method to reconstruct microglia, and quantify their features from 2D/3D image datasets. Multilevel thresholding is employed to segment soma volumes and recognize foreground voxels. Then, we propose a tracing process to connect seed points sampled from the foreground and form the skeleton of the branches. The thickness of the branches are estimated during a pruning process. The reconstructed data is then quantified and written in SWC standard file format. We have applied our methods to various image datasets for which the evaluated results show our methods outperform the state-of-the-art methods in terms of accuracy, precision, F1-measure, and computational time.
- Subject
- microscopy image; image processing; image modelling; image reconstruction; image analysis
- Identifier
- http://hdl.handle.net/1959.13/1410340
- Identifier
- uon:36169
- Rights
- Copyright 2019 Mahmoud Abdolhoseini
- Language
- eng
- Full Text
- Hits: 5296
- Visitors: 4637
- Downloads: 831
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | ATTACHMENT01 | Thesis | 14 MB | Adobe Acrobat PDF | View Details Download | ||
| View Details Download | ATTACHMENT02 | Abstract | 121 KB | Adobe Acrobat PDF | View Details Download |