Research Assistant
CurrentInherit differences in staining protocols and scanning hardware make digitized images in histopathology contain significant variability which is a limiting factor in creating image analysis algorithms which can be used for broad application across multiple institutions. We aim to address this limitation for image classifiers which classify patches of different tissue types taken from images of H&E stained colorectal cancer tissue. Automated analysis of characteristics of tissue samples including tumor depth, lymphocyte density, tumor budding, and co-localization of tumors buds with lymphatic and blood vessels has been shown to aid greatly in predicting patient outcomes and therefor can aid greatly in given the most appropriate treatment for patients.-Research has been presented at the Biomedical Engineering Society Conference and Image Network Ontario Symposium.-Work relies on matlab and python scripts
