A multi-process system for HEp-2 cells classification based on SVM

Abstract

This study addresses the classification problem of the HEp-2 cells using indirect immunofluorescence (IIF) image analysis, which can indicate the presence of autoimmune diseases by finding antibodies in the patient serum. Recently, studies have shown that it is possible to identify the cell patterns using IIF image analysis and machine learning techniques. In this paper we describe a system able to classify pre-segmented immunofluorescence images of HEp-2 cells into six classes. For this study we used the dataset provided for the participation to the "contest on performance evaluation on indirect immunofluorescence image analysis systems", hosted by the ICPR 2014. This system is based on multiple types of class-process and uses a two-level pyramid to retain some spatial information. We extract a large number (216) of features able to fully characterize the staining pattern of HEp-2 cells. We propose a classification approach based on the one-against-one (OAO) scheme. To do this, an ensemble of 15 support vector machines is used to classify each cell image. Leave-one-specimen-out cross validation method was used for the system optimization. The developed system was evaluated on a blind Hep-2 cells dataset performing a mean class accuracy (MCA) equal to 80.12%.