HEp-2 cells classification using locally aggregated features mapped in the dissimilarity space

Abstract

Indirect Immunofluorescence (IIF) followed by manual evaluation of the acquired slides from specialized personnel is the preferred laboratory technique used for the detection of Antinucleolar Antibodies (ANAs) in patient serum. In this procedure, several limitations appear and thus several automatic techniques have been proposed for the task of ANA detection. In this paper we propose a system for automatic classification of HEp-2 staining patterns, inspired by a recently proposed method for aggregating local image (SIFT) features into a compact and fixed length representation. More specifically we present a novel framework in which aggregated features are mapped into feature vectors in the dissimilarity space where the dimensionality of the descriptors is "naturally reduced". The final descriptor is low dimensional, while evaluation on a recently published dataset yields state of the art results.

HEp-2 staining pattern recognition at cell and specimen levels: Datasets, algorithms and results

“…From the connected components extracted from each of the so obtained binary images the following set of features is calculated: number of detected objects, density in binary image and mean objects solidity, complexity of cells contour. Texture is represented through both the distribution of the SIFT descriptor, calculated on a dense grid over the cell image and encoded according to the VLAD scheme as in Kastaniotis et al (2013), and the histograms of three variants of LBP, i.e. the uniform, the four patches and the center-symmetric ones.…”