Cellular outline segmentation using fractal estimators
Segmentation in biological images is essential for the determination of biological parameters that allow the construction of models of several biological problems. This helps to establish clear relationships between those models and the parameter estimation, and for elaboration of key experiments th...
|Autores principales:||, , , ,|
|Acceso en línea:||http://sedici.unlp.edu.ar/handle/10915/9536|
|Aporte de:||SEDICI (UNLP) de Universidad Nacional de La Plata Ver origen|
|Sumario:||Segmentation in biological images is essential for the determination of biological parameters that allow the construction of models of several biological problems. This helps to establish clear relationships between those models and the parameter estimation, and for elaboration of key experiments that give support to biological theories. Segmentation is the process of qualitative or quantitative information extraction (shape, texture, physical and geometric properties, among others). These quantities are needed to compute the biological descriptors for further classification (v.g., cell counting, development stage assessment, and many others). This process is almost always supervised (i.e., human assisted), since the quality of the images that are produced with classic microscopy technologies have defects that in general disallow the application of unsupervised segmentation techniques. In this paper we investigate the use of the a local fractal dimension estimation as an image descriptor for microscopy images. This local descriptor appears to be robust enough to perform unsupervised or semisupervised segmentations, specifically in our study. We applied this technique on microscopy images of amphibian embryos' skin in which, using immunofluorescence techniques, we have labeled the cell adhesion molecule E-Cadherin. This molecule is one of the key factors of the Ca<sup>2+</sup>- dependent cell-cell adhesion. Segmentation of the cellular outlines was performed using a processing workflow, which can be repeatedly applied to a set of similar images, from which information is extracted for characterization and eventual quantification purposes.|