Integration of Python-Based Software and UNet Machine Learning for Enhanced Bioimage Analytics
Additional Funding Sources
The project described was supported by the Center of Excellence in Biomedical Research through the Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health under Grant Nos. P20GM109095 and P20GM103408 and the National Science Foundation S-STEM Gateway Scholarships in Biological Sciences under Grant Award No. DUE-1644233.
Presentation Date
7-2022
Abstract
Our lab’s work revolves around understanding the relationship between the Linker of Nucleoskeletal and Cytoskeletal (LINC) complex and physical cell characteristics, with may provide a greater understanding of several tissue degenerative diseases related to the LINC complex. Bioimaging - where biology and image analysis meet - is a key part of data acquisition in this pursuit, specifically for quantifying cell vitality and immunofluorescence. However, while existing tools such as ImageJ, CellProfiler, and others can provide such functionalities, their high manual workload makes them prone to human error and time-consuming to use on large volumes of data. This project focuses on research and development of an original image analysis software tailored toward the needs of the lab - combining a multitude of cell data analysis capabilities with a greatly simplified workload for the user. Code was written in the Python language in a PyCharm environment, with a UNet machine learning model playing a critical role in proper image identification. The program is currently capable of analyzing both still and live-cell imaging files for cell count, movement, and immunofluorescent intensities for multiple channels. It is proving to be a major improvement over the previous programs used by our lab, providing a greater breadth of more accurate data that our lab can use to accelerate and better validate its findings at the microscopic level, and further its understanding of the LINC complex in a biomedical context.
Integration of Python-Based Software and UNet Machine Learning for Enhanced Bioimage Analytics
Our lab’s work revolves around understanding the relationship between the Linker of Nucleoskeletal and Cytoskeletal (LINC) complex and physical cell characteristics, with may provide a greater understanding of several tissue degenerative diseases related to the LINC complex. Bioimaging - where biology and image analysis meet - is a key part of data acquisition in this pursuit, specifically for quantifying cell vitality and immunofluorescence. However, while existing tools such as ImageJ, CellProfiler, and others can provide such functionalities, their high manual workload makes them prone to human error and time-consuming to use on large volumes of data. This project focuses on research and development of an original image analysis software tailored toward the needs of the lab - combining a multitude of cell data analysis capabilities with a greatly simplified workload for the user. Code was written in the Python language in a PyCharm environment, with a UNet machine learning model playing a critical role in proper image identification. The program is currently capable of analyzing both still and live-cell imaging files for cell count, movement, and immunofluorescent intensities for multiple channels. It is proving to be a major improvement over the previous programs used by our lab, providing a greater breadth of more accurate data that our lab can use to accelerate and better validate its findings at the microscopic level, and further its understanding of the LINC complex in a biomedical context.