Mercer, Joel (2014) Automated image analysis of estrogen receptor immunohistochemistry in breast cancer. Masters thesis, Memorial University of Newfoundland.
PDF (Migrated (PDF/A Conversion) from original format: (application/pdf))
- Accepted Version
Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.
Breast cancer is among the leading causes of cancer death in women. The estrogen receptor test (ERT) guides therapy and has a direct impact on patient survival. Recent data have shown the interpretation of ERT can vary both within and between laboratories. The aim of this project was to develop an automated system to classify digital images of ERTs as positive or negative. The slides were classified by two expert breast pathologists prior to being analyzed by the software. Two data sets were used for analysis, one from Eastern Health (Newfoundland) with 60 cases and the other from the UK with 275 cases. The UK dataset was also analyzed by two existing estrogen receptor analysis systems. The developed algorithm made use of a color deconvolution algorithm to separate the histological stains. It identified the individual cells by use of mathematical morphological operations and a marker based watershed algorithm. Finally, each cell was classified based on its concentrations of stains; if the proportion of positive cells reached a threshold then the case was deemed positive. Overall, the program reached a sensitivity and specificity of 100% for the first dataset. For the second data set 94% sensitivity and 95% specificity. This was an increase of 10% in sensitivity over the compared software. Given the history of false negatives with the ERT, an increase in sensitivty could offer increased patient safety with use of this software as a quality assurance system. Furthermore, its use could be extended to any immunohistochemical stain of any tissue, human or otherwise.
|Item Type:||Thesis (Masters)|
|Additional Information:||Includes bibliographical references (pages 81-87).|
|Department(s):||Engineering and Applied Science, Faculty of|
|Library of Congress Subject Heading:||Image processing--Digital techniques; Image processing -- Mathematics; Hormone receptors--Testing--Data processing; Immunohistochemistry--Technique; Biometry|
Actions (login required)