Abstract—The purpose of this paper is to assess the value of neural networks for classification of cancer and noncancer prostate cells. Gauss Markov Random Fields, Fourier entropy and wavelet average deviation features are calculated from 80 noncancer and 80 cancer prostate cell nuclei. For classification, artificial neural network techniques which are multilayer perceptron, radial basis function and learning vector quantization are used. Two methods are utilized for multilayer perceptron. First method has single hidden layer and between 3-15 nodes, second method has two hidden layer and each layer has between 3-15 nodes. Overall classification rate of 86.88% is achieved. Keywords—Artificial neural networks, texture classification, cancer diagnosis. I. INTRODUCTION ANCER is one of the biggest problems of the human beings and in many developed countries, prostate cancer is one of the commonly diagnosed cancer in men. Risk factor of the prostate cancer depends on age, genetic background, and ethnical character. Diagnosis of prostate cancer requires the tissue and cell specimens. These specimens (as shown in Fig.1) are screened and analyzed by a pathologist using a microscope. Optimum medical treatment is decided according to this information gathered by the pathologist. In some cases, correct diagnosis is very hard and there can be 30-40% difference between pathologists’ decisions [1]. Dramatic results about wrong diagnosis of cancer cases from biopsy slides can be found in [2]. Prostate cancer is evaluated using to staging systems: the Jewett-Whitmore system and the TNM (tumor, node, metastases) system. In the Jewett-Whitmore system, prostate cancer is classified according to anatomical view and spread. A and B are early stages. In these stages there are few cancerous cells and they are in the prostate tissue. In the later stages (C and D) cancer invades most of the prostate and This work is funded by Dokuz Eylül Univ., BAP under grant 02.KB.FEN.058. M. Makinacı is with the Electrical and Electronics Engineering Department, Dokuz Eylül University, İzmir, Turkey (phone: +90 232-412- 7160; fax: +90 232-453-1085; e-mail: makinaci@eee.deu.edu.tr). M. Sinecen is with the Electrical and Electronics Engineering Department, Dokuz Eylül University, İzmir, Turkey (e-mail: msinecen@eee.deu.edu.tr). spreads to other organs/tissues. In TNM system, T refers to the size of the primary tumor, N will describe the extent of lymph node involvement, and M refers to the presence or absence of metastases. Artificial neural networks are commonly used for the diagnosis of the prostate cancer. In these studies, various types of data are used, such as prostate specific antigen (PSA) levels [3], clinical and biochemical criteria [4], ultrasonic echo signals with PSA [5]. Neural networks are utilized in this kind of biomedical applications because of their ability to perform more accurately than other classification techniques. Basic advantages of the neural network method over traditional classifiers are; easy adaptation to different types of data and use of its complex configuration to find the best nonlinear function between the input and the output data. II. DATA COLLECTION A. Image Acquisition The specimen images are x100 magnified by the Leica microscope. An oil immersion objective was used. The analog image signal was acquired with a color camera and with a s- video connection the signal was transmitted to the computer. The images are digitalized in 768x576 pixel 24bit/pixel format and saved. Fig. 1 Prostate tissue containing cancerous cell nuclei M. Sinecen, and M. Makinacı Classification of Prostate Cell Nuclei using Artificial Neural Network Methods C World Academy of Science, Engineering and Technology International Journal of Medical and Health Sciences Vol:1, No:7, 2007 474 International Scholarly and Scientific Research & Innovation 1(7) 2007 scholar.waset.org/1307-6892/14915 International Science Index, Medical and Health Sciences Vol:1, No:7, 2007 waset.org/Publication/14915