Automatic Classification of HEp-2 Cells Using Multi-dimensional Local Binary Patterns

Indirect immunofluorescence imaging is a fundamental technique used for detecting antinuclear antibodies in HEp-2 cells. This is particularly useful for the diagnosis of autoimmune diseases and other important pathological conditions involving the immune system. HEp-2 cells can be categorised into six groups: homogeneous, fine speckled, coarse speckled, nucleolar, cytoplasmic, and Centro mere cells, which give indications of different autoimmune diseases. This categorisation is typically performed through manual evaluation which is time consuming and subjective. In this paper, we propose a method for automatic classification of HEp-2 cells using local binary pattern (LBP) based texture descriptors. LBP is a simple yet powerful texture algorithm which encodes the relationship of pixels to their local neighbourhood. In our approach, we utilise multi-dimensional LBP (MD-LBP) histograms, which record multi-scale texture information while maintaining the relationships between the scales, and perform classification using support vector machines. We demonstrate our algorithm to work well on a dataset of 721 cell images, giving a correct classification rate exceeding 95%, which is particularly impressive as it is based solely on a single type of image feature.