Gillian Holder, “Segmentation Methods of for Tumor Tissues”
Mentor: Mahsa Dabagh, Biomedical Engineering
Poster #60

With new advances in machine learning (ML), digital histology can be made easier and more accurate while reducing clinician’s workload to diagnose and detect cancer. For ML codes to be trained to identify key components of tumor tissue, the tissue must be prepared and stained with the right color and intensity to be successfully analyzed. The common hematoxylin-eosin stain is used to stain nucleic acids and proteins in tissue. Well-fixed cells show considerable intranuclear detail, as well. On the other hand, picrosirius red hematoxylin can clearly segment the stomal boundaries and decomposition color. In this review, these tissue staining techniques are researched and studied where ML can be integrated to identify tumor tissues.