Urine sediment image segmentation based on feedforward backpropagation neural network

The appearance of crystals, casts, red blood cells, white blood cells and bacteria or yeast in urine sediment is a major clinical significance. It provides important information for both diagnosis and prognosis. However, low contrast against the background, less illuminating environment and an existent of complicated components on the microscopic urine sediment image need more sophisticated method to analyze. In this paper, we present a conventional method to segment the urine-sediment visual component by using feedforward-backpropagation algorithm of neural network. Background color was used as a main feature in the segmentation process. Experimental result shows that our proposed method provides quite satisfactory segmentation.