Nuclei identification is a pivotal first step in many areas of biomedical research. Pathologists often observe images containing microscopic nuclei as part of their day to day jobs. During research, pathologists must identify nuclei characteristics from microscopic images such as: volume of nuclei, size, density and individual position within image. The pathology field can benefit from image detection enhancements done through the use of computer image segmentation techniques. This research presents methods that can be used to identify all the cell nuclei contained in images. Multiple techniques were experimented with such as edge detection and Convolutional Neural Networks with U-Net architecture. The data for training these models was sourced from the 2018 Data Science Bowl sponsored by Kaggle and Booz, Allen, Hamilton. As a result, there were various methods identified to assist the pathology industry for automating nuclei detection by using computer image detection methods. These computer methods rapidly process images for research purposes, with a reasonably high accuracy which has the potential to greatly accelerate the pace of research.
Flores, Jonathan; Prasad, Thejas; Kassof, Jordan; and Slater, Robert
"Automate Nuclei Detection Using Neural Networks,"
SMU Data Science Review: Vol. 2
, Article 8.
Available at: https://scholar.smu.edu/datasciencereview/vol2/iss1/8
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License