HEp-2 cell classification in indirect immunofluorescence images

Indirect immunofluorescence (IIF) with HEp-2 cells has been used to detect antinuclear auto-antibodies (ANA) for diagnosing systemic autoimmune diseases. The aim of this study is to develop an automatic scheme to identify the fluorescence pattern of HEp-2 cell in the IIF images. By using the previously proposed two-staged segmentation method, the similarity-based watershed algorithm with marker techniques was performed to segment each fluorescence cell. Then the proposed classification method utilized learning vector quantization (LVQ) with eight textural features to identify the fluorescence pattern. This study evaluated 1036 autoantibody fluorescence patterns from 44 IIF images that were divided into six pattern categories (including diffuse, peripheral, coarse speckled, fine speckled, discrete speckled and nucleolar patterns). The simulations show that the proposed system differentiates autoantibody fluorescence patterns with a good result and is therefore clinically useful to provide a second opinion.