AI-Based Image Analysis on Bone Mineralization
Bone mineralization by osteoblast is a complex and dynamic process. Osteoblasts mineralize the osteoid, facilitating the nucleation and growth of minerals. Once fully mineralized, osteoblasts enter a resting state lining cells until the next cycle of bone formation. While this cellular mechanism is well known, how the underlying collagen structure controls mineral deposition is still being actively researched. Research on this topic requires extensive image analysis and tracking of mineralization but current methods like thresholding may limit the yield of accuracy of area as well as proper visualization. Here, I used NiS elements to train an AI to track and segment mineralization on images in hopes of accurately recording the area of mineralization over time and providing clearer visualization of the changes in mineralization. First, I annotated the mineralized matrix as opposed to the demineralized matrix with binaries on varying images. The desired set of annotated images is used to train an AI where the AI can then learn to distinguish mineralization from demineralization based on the annotated binaries. A GA3 pipeline is created for the AI to segment the images and then the necessary image-processing functions are added to improve the quality and accuracy of the segmentations and record the area of segmented objects. The data is then graphed and a video of the segmented mineralization is created. These results precisely portray the change in mineralization over time and validate the use of this image analysis for future research in bone homeostasis and disease.
Research Area | Presenter | Title | Keywords |
---|---|---|---|
Cancer Studies | Rasku-Casas, Isabella | Image Analysis | |
Medical Sciences | Nguyen, Vy | Biomineralization | |
Biological Organisms | Portillo, Neida O. | microscopy |