Open Collections

Methods for improved imaging and analysis of tissue-based biomarkers

University of British Columbia

Study of molecular biomarkers can provide insight into the molecular complexity of cancer, create new cancer screening tools, monitor treatment’s effectiveness, and predict patient’s response to treatment. This thesis proposes novel methods for the improved quantitative analysis of labeled molecular biomarkers in tissue sections. This is a necessary step towards the ultimate goal of personalized treatment of cancer. As 85% of all cancers arise in epithelial tissue, we have developed means for objectively and quantitatively assessing the distribution of a molecular biomarker in epithelial tissue sections. We have applied this means to characterize the spatial distribution of proliferating cells in 613 normal and pre-neoplastic bronchial epithelial biopsies. We have demonstrated, for the first time ever, that the knowledge of the spatial distribution of proliferating cells enables prediction of the outcome of lung intraepithelial lesions. We have developed methods for the automated and quantitative assessment of the expression of tissue-based molecular biomarkers on a cell-by-cell basis. This is achieved by multispectral imaging of labeled tissue sections. We have proposed methods for unsupervised linear spectral unmixing of multispectral images for the purpose of identifying individual labels in a multiple labeled tissue section. We have demonstrated that the use of multispectral imaging combined with our proposed analysis methods quantitatively improves the results of cell nuclei identification compared to three-color RGB imaging, in more than 22,000 cells in 58 tissue sections with nuclear, cytoplasmic, or membrane bound biomarkers. Finally, we have developed an imaging method for capturing images representing biomarkers in a tissue. Compared to multispectral imaging, our proposed imaging method significantly reduces the number of captured images required for the identification of a biomarker in a tissue. This method uses images captured under a series of narrow-band illumination spectra φ_i,i=1,2,…,N to find a weighted linear combination of the images that represents a certain component in a tissue. The weights in the weighted linear combination of images are then used to design one or two illumination spectra as weighted linear combinations of the narrow-band spectra φ_i. Images representing the component of interest are then captured under the designed illuminations.

Text

2012-12-20

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/43728

Electrical and Computer Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/