R. Khosla et al. (Eds.): KES 2005, LNAI 3682, pp. 123130, 2005. ' Springer-Verlag Berlin Heidelberg 2005 Performance Comparison for MLP Networks Using Various Back Propagation Algorithms for Breast Cancer Diagnosis S. Esugasini 1 , Mohd Yusoff Mashor 1 , Nor Ashidi Mat Isa 1 , and Nor Hayati Othman 2 1 Control and Electronic Intelligent System (CELIS) Research Group School of Electrical & Electronic Engineering, Universiti Sains Malaysia Engineering Campus, 14300 Nibong Tebal, Penang Malaysia esu@hotmail.com, {yusof,ashidi}@eng.usm.my 2 School Medical Sciences, Universiti Sains Malaysia, Health Campus 16150 Kubang Kerian, Kelantan, Malaysia Abstract. This paper represents the performance comparison of the Multilay- ered Perceptron (MLP) networks using various back propagation (BP) algo- rithms for breast cancer diagnosis. The training algorithms used are gradient descent with momentum and adaptive learning, resilient back propagation, Quasi-Newton and Levenberg-Marquardt. The performances of these four algo- rithms are compared with the standard steepest descent back propagation algo- rithm. The current study investigates and compares the accuracy, sensitivity, specificity, false negative and false positive results of the selected four algo- rithms to train MLP networks. The Papinicolou image of breast cancer cells were captured via an image analyzer and thirteen morphological features were extracted to numerical scores. The feature scores are used as data sets to train the MLP network. The MLP network using the Levenberg-Marquardt algorithm displays the best performance for all the five measurement criterias (accuracy, specificity, sensitivity, true positive and true negative) at a lower number of hidden nodes. 1 Introduction A total of 26,089 patients were diagnosed with cancer disease among all residents in Peninsular Malaysia, in the year 2002, comprising 54% of the amount to be female. The crude rate in the same year was 148.4 per 100,000 populations for females, with the breast cancer being the number one killer. Breast cancer accounted 30.4% of newly diagnosed cancer cases in Malaysian women, with the possibility of 1 of every 19 woman in Malaysia has the risk to develop breast cancer in their lifetime [1]. Sev- eral risk factors have been identified by researchers over the year. Age and genetic are the major factors related to breast cancer. The women at risk age vary from 35 to 55 and women with first degree relatives with the history of breast cancer are at higher risk than the rest of the population. An organized mass screening is not yet practiced in Malaysia although the number of cases is increasing each year. The key factor for this situation is the number of specialist (pathology, etc) required for mass screening is very less in Malaysia. Malaysian Ministry of Health, 1999 states that the number of specialist to population ratio is about 1:20 000 in Malaysia [2].