Immunohistochemistry image analysis : protein, nuclei and gland
Publication Date
December 9, 2014
Creator
Abstract
This thesis focus on the analysis of digitized microscopic image, especially on IHC stained colour images. The corresponding contributions focused on the automatic detection of stain colour and glands, the segmentation and quantification of cell nuclei, the analysis of liver cirrhosis and the development of a semi-automatic toolbox.
Colour is the most important feature in the analysis of immunostained images. We developed a statistical colour detection model for stain colour detection based on the histograms of collected colour pixels. This is acting on the approach "what you see is what you get" which outperforms the other methods on the detection of several kinds of stain colour.
Verifying the presence of nuclei and quantifying positive nuclei is the foundation of cancer grading. We developed a novel seeded nuclei segmentation method which greatly improves the segmentation accuracy and reduces both over-segmentation and under-segmentation. This method has been demonstrated to be robust and accurate in both segmentation and quantification against manual labelling and counting in the evaluation process.
The analysis of gland architecture, which reflects the cancer stage, has evolved into an important aspect of cancer detection. A novel morphology-based approach has been developed to segment gland structures in H-DAB stained images. This method locates the gland by focusing on its morphology and intensity characteristics, which covers variations in stain colours in different IHC images. The evaluation results have demonstrated the improvements of accuracy and efficiency.
For the successive development of three methods, we put them in a semi-automatic toolbox for the aid of IHC image analysis. It can detect different kinds of stain colour and the basic components in an IHC image. The user created models and parameters can be saved and transferred to different users for the reproduction of detection results in different laboratories.
To demonstrate the flexibility of our developed stained colour detection technique, the tool has been extended to the analysis of liver cirrhosis. It is a novel method based on our statistical colour detection model which greatly improves the analysis accuracy and reduces the time cost.
Item Type
ethesis
Thesis Type
PhD
Subjects (LC)
Associated Schools / Departments
School of Computer Science (UK)
eprints ID
27616
UoN Repository URI
Except where otherwise noted, this item's license is described as
File(s)![Thumbnail Image]()
Name
Thesis.pdf
Type
Full-text
Size
11.31 MB
Format
Adobe PDF
Checksum (MD5)
862c6abbad3cde649d0f06699e716a9d