NUCLEI SHAPE ANALYSIS, A STATISTICAL APPROACH
Abstract
The method presented in our paper suggests the use of Functional Data Analysis (FDA) techniques in an attempt to characterise the nuclei of two types of cells: Cancer and non-cancer, based on their 2 dimensional profiles. The characteristics of the profile itself, as traced by its X and Y coordinates, their first and second derivatives, their variability and use in characterization are the main focus of this approach which is not constrained to star shaped nuclei. Findings: Principal components created from the coordinates relate to shape with significant differences between nuclei type. Characterisations for each type of profile were found.
Keywords
2-dimensional profile; cancer detection; functional data analysis; principal differential analysis;star shape
DOI: 10.5566/ias.v27.p1-10
Copyright (c) 2014 Image Analysis & Stereology