City Research Online

Abstract

This work analyses breast cancer histopathology images with the objective of performing segmentation of the different tissue components, with special focus on the tumour region, which is of irregular shape and of complex characteristics, in terms of texture, light intensity, and contrast. On the other hand, the images come from specialized slide scanners, that digitises the glass slide into a high resolution digital file. Manipulating a single gigabyte-scale image is a challenging and time-consuming task, requiring an efficient implementation during experimentation. The images analysed are primarily of breast cancer, although the selected approach was also used for prostate cancer images. Two approaches have been explored for this research work, a hand-craft method based on image processing algorithms for feature extraction, and Deep Learning models. Both methods reveal the evolution in digital pathology during the last decade, and both were analysed and implemented in different clinical tasks. The first task is the tumour cellularity assessment in patients under Neo-Adjuvant treatment, which is a medical treatment of breast cancer given before the main treatment. In this experiment, the tumour area of the image is estimated. Next, multi-class segmentation is analysed on breast cancer tissue, with the main goal of segmenting the different tissue components. All of the images come from the grand challenge platform, and the results presented in this work were also submitted to the corresponding challenge. This thesis provides a comprehensive workflow of the digital pathology for breast cancer analysis, covering both approaches. Future research can be done in other types of cancer as well as in newer pathology stained images.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics > QA76 Computer software Q Science > QM Human anatomy R Medicine > RC Internal medicine R Medicine > RC Internal medicine > RC0254 Neoplasms. Tumors. Oncology (including Cancer) T Technology > T Technology (General)
Departments:	School of Science & Technology > Computer Science School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads