Rough Sets and Local Texture Features for Diagnosis of Connective Tissue Disorders

The standard method for diagnosis of connective tissue disorders is based on the automatic classification of antinuclear autoantibodies by analyzing indirect immunofluorescence images of human epithelial type 2 (HEp-2) cells. In this regard, the paper presents a new method to select relevant texture features for HEp-2 cell staining pattern recognition. The proposed method is developed by judiciously integrating the theory of rough sets and the merits of local texture descriptors. While hypercuboid equivalence partition matrix of rough sets helps to select important texture descriptors for HEp-2 cell classification, the maximum relevance-maximum significance criterion of feature selection facilitates identification of significant and relevant features under important descriptors. Finally, support vector machine with different kernels as well as extreme learning machine are used to recognize one of the known staining patterns present in HEp-2 cell images. The effectiveness of the proposed method, along with a comparison with related approaches, is demonstrated on publicly available MIVIA HEp-2 cell image database.