Recognition of early phase of atherosclerosis using principles component analysis and artificial neural networks from carotid artery Doppler signals Fatma Dirgenali a , Sadik Kara b, * a Turkish Standards Institution, Organized Industry Area, 6.street, 38512 Kayseri, Turkey b Department of Electrical-Electronics Engineering, Erciyes University, 38039 Kayseri, Turkey Abstract Atherosclerosis means thickening and hardening of the arteries, which has dramatic effects on blood pressure, resistance and blood flow. Since angiography is invasive and has a relatively high cost, non-invasive ultrasonic Doppler sonography is generally recommended to diagnose of athersosclerosis. In this study, we have employed the sonograms depicted from Autoregressive (AR) modeling, Principles component analysis (PCA) for data reduction of Doppler sonograms and artificial neural networks (ANN) in order to distinguish between atherosclerosis and healthy subjects. The fuzzy appearance of the carotid artery Doppler signals makes physicians suspicious about the existence of diseases and causes false diagnosis. Our technique gets around this problem using ANN to decide and assist the physician to make the final judgment in confidence. The stated results show that training time and processing complexity were reduced using PCA-ANN architecture however the proposed method can make an effective interpretation and ANN classified Doppler signals successfully. q 2005 Elsevier Ltd. All rights reserved. Keywords: Atherosclerosis; Carotid artery; Doppler signals; Autoregressive modelling; Principles component analysis; Artificial neural networks 1. Introduction Atherosclerosis means thickening and hardening of the arteries. It usually affects large and medium-sized arteries. It is characterized by calcified plaques, lipids, and cellular debris in the inner layers of the walls of large and medium-sized arteries. Plaques can grow large enough to significantly reduce the blood’s flow through an artery. Decreased aortic compliance is associated with atherosclerotic events (Hirai, Sasayama, Kawasaki, & Yagi, 1989; Stefanadis, Stratos, Boudoulas, Kourouklis & Toutouzas, 1990; Dart et al., 1991) and also with increasing age (Avolio, Chen, & Wang, 1983), hypertension (Heints, Gillessen, & Walkenhorst, 1993; Liu, Ting, & Zhu, 1989), diabetes (Airaksinen, Salmela, & Linnaluoto, 1993), hyperlipidemia (Lehmann, Watts, & Gosling, 1992), and smoking (Stefanadis, Tsiamis, & Vlachopoulos, 1997). These findings suggest the potential utility of decreased aortic compliance as a cardiovascular risk factor, which is predictive of cardiovascular mortality (Syeda et al., 2003). When atherosclerosis develops in the arteries that supply the brain (carotid arteries), a stroke may occur; when it develops in the arteries that supply the heart (coronary arteries), a heart attack may occur. The narrowing and hardening of the arteries has dramatic effects on blood pressure, resistance and blood flow. Resistance increases when radius decreases, as friction of blood flow against vessel wall increases. Therefore the circulation of blood flow is reduced. Development of a plaque also deforms the endothelial wall, increasing turbulent flow and increasing resistance. The hardening of the arterial walls, increases resistance to flow (Hoskins, McDicken, & Allan, 2000; Libley, 2001). There are a number of tests that of arterial compliance, including blood tests, coronary angiography and ultrasound. Invasive, sophisticated clinical measurements have provided data from recordings of arterial blood flow, pressure, and diameter changes. When the symptoms develop, catheter angiography is considered as the gold standard to detect and quantify the stenosis. Since angiography is invasive and has a relatively high cost, non-invasive ultrasonic Doppler sono- graphy is generally recommended. Recent advances in the Doppler imaging technique have made possible evaluation of the temporal and spatial flow characteristics in the different Expert Systems with Applications 31 (2006) 643–651 www.elsevier.com/locate/eswa 0957-4174/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2005.09.064 * Corresponding author. Address: Electronics Engineering Department, Biomedical Engineering Group, Erciyes University, 38039 Kayseri, Turkey. Tel.: C90 352 4374901x32228; fax: C90 352 4375784. E-mail addresses: fdirgenali@tse.org.tr (F. Dirgenali), kara@erciyes.edu.tr (S. Kara).