Neurofuzzy classification of the effect of diabetes mellitus on carotid artery

Abstract

Here, we present a new clasification system for the effects of diabetes mellitus (DM) on carotid artery by using neurofuzzy system. Blood flow hemodynamics were obtained from 118 carotid arteries of 59 patients with DM by using 7.5 MHz transducer and Doppler-M unit. Vmax = peak systole, Vmin = end diastole, resistive index (RI), and pulsatile index (PI) were measured with the use of Doppler sonography. These values were applied to neurofuzzy system using NEFCLASS model. With the increase of epoch from 200 to 500 and pruning of fuzzy rules, our classification system was found to be successful in 85% of the cases. In 100 of 118 patients the classificitaion system was found to be correct. Our findings suggest that with the application of Doppler signal parameters from carotid arteries to neurofuzzy system may produce a new and reliable classification system for diagnosing diameter stenosis.