A fuzzy model for the classification of abnormal substances in urinary sediment images

Abstract

A new model for recognition of abnormal substances in urinary sediment images (USI) is presented. Here the scheme has a particular focus on abnormal substances, whose shapes and sizes are not normal and the data of images are imperfection and uncertain. These substances are very difficult to be recognized. Our proposed method combines fuzzy computation with the statistical method for recognition. A database is created for storing the different types of substances which include normal and abnormal substances used as standard patterns. Then the similarity degrees of textures between the standard patterns and tested substances are calculated. Furthermore, “If-then” knowledge base is created on the basis of the experiences of human-expert. Finally, the knowledge base for each substance is applied to evaluate each abnormal substance by using fuzzy reasoning of the combination of texture similarities and other features, the precision of classification is improved. This method also provides a flexible way to improve the ability of recognition of substances.