|About this Abstract
||2018 TMS Annual Meeting & Exhibition
||Biological Materials Science
||Automatic Shape-based Cell Identification in Arabidopsis Thaliana Cotyledons Using 3D Moment Invariants
||Ryan Harrison, Marc De Graef
|On-Site Speaker (Planned)
The automated classification of cells is a powerful tool in biological analysis and has been a focus of study for fifty years. Previous work has primarily focused on the analysis of two-dimensional images, due to the prevalence of this type of data and the ease of its acquisition. Recent advances in staining and microscopy techniques have allowed for improved accuracy and precision in the non-destructive gathering of 3D cell data. In this work, the three-dimensional shapes of the cells of Arabidopsis thaliana cotyledons are quantified by different sets of 3D moment invariants. For each feature set, a cell is represented as a point in a high-dimensional space. By visualizing these spaces in two-dimensions, it is found that specific classes of cells create connected structures that are isolated using a hierarchical clustering algorithm. This technique allows for differentiation between types of cells without the training of a classifier.
||Planned: Supplemental Proceedings volume