Use of cytomorphometry for classification of subcellular patterns in 3D images

Abstract

This paper presents a methodology for the classification of subcellular patterns by the extraction of cytomorphometric features in 3D isosurfaces. In order to validate the proposal, we used a database of 3D images of HeLa cells with nine classes. For each cell, several morphological attributes were extracted based on its isosurface. Using the Quadratic Discriminant Analysis (QDA) classifier with the hybrid attribute selector, we achieved 97.59 of accuracy and F1-score of 0.9757 when classifying the subcellular patterns.