Recognition and Identification of Red Blood Cell Size Using Angular Radial Transform and Neural Networks

Abstract

In this paper, a novel method for the estimation of the human Red Blood Cell (RBC) size using light scattering images is presented. The information retrieval process includes, image normalization and features extraction using the Angular Radial Transform (ART). A Radial Basis Neural Network (RBF-NN) estimates the RBC geometrical properties. The proposed method is evaluated in both regression and identification tasks when three important geometrical properties of the human RBC are estimated using a database of 1575 simulated images generated with the boundary element method. The experimental setup consists of a light beam at 632.8 nm and moving RBCs in a thin glass and additive noise distortion is simulated using white Gaussian noise from 60 to 10 dB SNR. The regression and identification accuracy of actual RBC sizes is estimated using three feature sets, giving a mean error rate less than 1 percent of the actual RBC size, in case of noisy image data at 10 dB SNR or better, and more than 97 percent mean identification rate.