Automatic detection of red blood cells in hematological images using polar transformation and run-length matrix

Many of diseases related to red blood cells can be diagnosed by analyzing the hematological images. The first step is to locate precisely the position of red blood cells. In this paper, a novel method based on polar transformation and run-length matrix is proposed for detecting red blood cells in hematological images. The multilayer perceptron was employed for classifying the feature vectors. This technique detected 3022 of the 3092 red blood cells existed in the images studied. The results illustrate capability of the proposed method for detecting red blood cells in hematological images.