Automated quantitative and qualitative analysis of neuroblastoma cancer tissue

Tafavogh, S

Automated quantitative and qualitative analysis of neuroblastoma cancer tissue

Tafavogh, S

Permalink

Publication Type:: Thesis
Issue Date:: 2014

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download contents and abstractAdobe PDF (173.97 kB)

Adobe PDF

Download thesisAdobe PDF (3.28 MB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Tafavogh, S
dc.date.accessioned	2014-08-27T08:27:18Z
dc.date.available	2014-08-27T08:27:18Z
dc.date.issued	2014
dc.identifier.uri	http://hdl.handle.net/10453/29235
dc.description	University of Technology, Sydney. Faculty of Engineering and Information Technology.	en_US
dc.description.abstract	The goal of this thesis is to develop an innovative Computer Aided Diagnosis (CAD) system for the common deadly infant cancer of Neuroblastoma. Neuroblastoma accounts for more than 15% of childhood cancer deaths, and it has the lowest survival rate among the paediatric cancers in Australia. In quantitative analysis the total number of different regions of interest are counted, and qualitative analysis determines abnormalities within the tumour. Quantitative and qualitative analysis of tumor samples under the microscope is one of the key markers used by pathologists to determine the aggressiveness of the cancer, and consequently its therapy. Because of the variety of the histological region types and histological structures in the tissue, analyzing them under the microscope is a tedious and error-prone task for pathologists. The negative effects of inaccurate quantitative and qualitative analysis have led to an urgent call from pathologists for accurate, consistent and automated approaches. Computer Aided Diagnosis (CAD) is an automated cancer diagnostic and prognostic system which enhances the ability of pathologists in the quantitative and qualitative analysis of tumor tissues. However, there are four main issues with developing a CAD system for pathology labs: First is the fluctuating quality of the histological images. Second is a wide range of different types of histological regions and histological structures with complex morphology each adopting a specific algorithm. Third is overlapping cells which decrease the accuracy of quantitative analysis. Fourth is a lack of utility for pathology labs when they do not follow an appropriate clinical prognosis scheme. Moreover, most of the proposed CAD systems perform either quantitative or quantitative analysis and only very few of them manipulate both types of analysis on the cancerous tumor tissue. This thesis aims to address the issues raised by developing an innovative CAD system that assists pathologists in determining a more appropriate prognosis for the leading infant cancer of Neuroblastoma. The CAD will automatically perform quantitative and qualitative analysis on images of tumor tissue to extract specific histological regions and histological structures which are used for determining the prognosis for Neuroblastoma. This thesis has four main contributions. Contribution 1 develops novel algorithms to enhance the quality of histological images by reducing the wide range of intensity variations. Contribution 2 proposes a series of segmentation algorithms for extracting different types of histological regions and histological structures. Contribution 3 addresses the issue of overlapping cells by developing algorithms for splitting them into single cells. Contribution 4 grades the aggressiveness level of neuroblastoma tumor by developing a prognosis decision engine. The main outcomes of the proposed CAD system in this thesis are a series of novel algorithms for enhancing the quality of the histological images and for segmenting histological regions and histological structures of interests, introducing a prognosis decision engine for grading a neuroblastoma tumor based on a well established histopathological scheme, facilitating the process of prognosis and tumor classification by performing accurate and consistent quantitative and qualitative tissue analysis, and enhancing digital pathology by incorporating a digital and automated system in the work flow of pathologists. The performance of all the developed algorithms in this thesis in terms of correctly extracting histological regions, histological structures and grading the level of tumor aggressiveness, is evaluated by a pathologist from the department of histopathology in the Children's Hospital at Westmead, Sydney. Moreover, all the results are compared with state of the art methods. The results indicate that the algorithms proposed in this thesis outperform state of the art quantitative and qualitative methods of analysis.	en_US
dc.format	Thesis (PhD)	en_US
dc.language.iso	en	en_US
dc.relation	https://opus.lib.uts.edu.au/bitstream/10453/29235/2/02whole.pdf
dc.rights	info:eu-repo/semantics/openAccess
dc.rights	The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.
dc.rights	au.edu.uts.lib/ppc
dc.subject	Neuroblastoma.	en
dc.subject	Computer Aided Diagnosis (CAD)	en
dc.subject	Children.	en
dc.subject	Cancer.	en
dc.title	Automated quantitative and qualitative analysis of neuroblastoma cancer tissue	en_US
dc.type	Thesis
utslib.copyright.status	open_access

Abstract:

The goal of this thesis is to develop an innovative Computer Aided Diagnosis (CAD) system for the common deadly infant cancer of Neuroblastoma. Neuroblastoma accounts for more than 15% of childhood cancer deaths, and it has the lowest survival rate among the paediatric cancers in Australia. In quantitative analysis the total number of different regions of interest are counted, and qualitative analysis determines abnormalities within the tumour. Quantitative and qualitative analysis of tumor samples under the microscope is one of the key markers used by pathologists to determine the aggressiveness of the cancer, and consequently its therapy. Because of the variety of the histological region types and histological structures in the tissue, analyzing them under the microscope is a tedious and error-prone task for pathologists. The negative effects of inaccurate quantitative and qualitative analysis have led to an urgent call from pathologists for accurate, consistent and automated approaches. Computer Aided Diagnosis (CAD) is an automated cancer diagnostic and prognostic system which enhances the ability of pathologists in the quantitative and qualitative analysis of tumor tissues. However, there are four main issues with developing a CAD system for pathology labs: First is the fluctuating quality of the histological images. Second is a wide range of different types of histological regions and histological structures with complex morphology each adopting a specific algorithm. Third is overlapping cells which decrease the accuracy of quantitative analysis. Fourth is a lack of utility for pathology labs when they do not follow an appropriate clinical prognosis scheme. Moreover, most of the proposed CAD systems perform either quantitative or quantitative analysis and only very few of them manipulate both types of analysis on the cancerous tumor tissue. This thesis aims to address the issues raised by developing an innovative CAD system that assists pathologists in determining a more appropriate prognosis for the leading infant cancer of Neuroblastoma. The CAD will automatically perform quantitative and qualitative analysis on images of tumor tissue to extract specific histological regions and histological structures which are used for determining the prognosis for Neuroblastoma. This thesis has four main contributions. Contribution 1 develops novel algorithms to enhance the quality of histological images by reducing the wide range of intensity variations. Contribution 2 proposes a series of segmentation algorithms for extracting different types of histological regions and histological structures. Contribution 3 addresses the issue of overlapping cells by developing algorithms for splitting them into single cells. Contribution 4 grades the aggressiveness level of neuroblastoma tumor by developing a prognosis decision engine. The main outcomes of the proposed CAD system in this thesis are a series of novel algorithms for enhancing the quality of the histological images and for segmenting histological regions and histological structures of interests, introducing a prognosis decision engine for grading a neuroblastoma tumor based on a well established histopathological scheme, facilitating the process of prognosis and tumor classification by performing accurate and consistent quantitative and qualitative tissue analysis, and enhancing digital pathology by incorporating a digital and automated system in the work flow of pathologists. The performance of all the developed algorithms in this thesis in terms of correctly extracting histological regions, histological structures and grading the level of tumor aggressiveness, is evaluated by a pathologist from the department of histopathology in the Children's Hospital at Westmead, Sydney. Moreover, all the results are compared with state of the art methods. The results indicate that the algorithms proposed in this thesis outperform state of the art quantitative and qualitative methods of analysis.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/29235