Main content

A Perceptual Colour Separation Methodology for Automated Quantification of Ki67 and Hematoxylin Stained Digital Histopathology Images

Show full item record

Title: A Perceptual Colour Separation Methodology for Automated Quantification of Ki67 and Hematoxylin Stained Digital Histopathology Images
Author: Morreale, Peter
Department: School of Engineering
Program: Engineering
Advisor: Dony, Robert
Abstract: This thesis is focussed on the development of a colour separation methodology for quantification of histopathology stains Ki67 and hematoxylin. Traditional methods require tedious manual evaluations prone to inter and intra-observer variability due to inherent subjectivity. Automatic algorithms have been proposed as the solution to manual challenges, however many algorithms are sensitive to wide staining variability common among histopathology images. This thesis proposes a perceptual colour separation framework that models colour information in a way that mimics the human visual systems ability to identify colour content. It is an automatic framework that was evaluated against 30 manually labelled canine mammary TMA core images. The average difference between manual and automatic PI estimates was 3.25%, with a Kappa similarity statistic of 0.86 across four Ki67 cut-off levels, and a linear correlation coefficient of 0.93. Validation studies show the colour separation framework achieves robust results across various Ki67 staining levels.
URI: http://hdl.handle.net/10214/13021
Date: 2018-04
Terms of Use: All items in the Atrium are protected by copyright with all rights reserved unless otherwise indicated.


Files in this item

Files Size Format View
Morreale_Peter_201804_Masc.pdf 4.460Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record