Presentation of a New Method for the Classification of Angiography Results Using Artificial Neural Networks

Researches have shown that angiography is a more acceptable diagnostic tool for coronary artery diseases. In the present study, angiographic results were classified by artificial neural networks to determine whether coronary arteries are closed or not. The present study was performed at Kowsar Hospital in Shiraz in 2013. The study population consisted of individuals undergoing coronary artery angiography in August 2013; in total, 152 participants were randomly selected. In this study, multilayer perceptron (MLP) and probabilistic neural network (PNN) were utilized to classify subjects into two groups: patients and healthy participants. For network design, 85% of the data was used for network training and 15% for testing the network. For implementing the network, MATLAB software version 7.12.0 was used. The simulations were made using a 2.4 GHz Corei 5 system with 4 GB memory (Windows 7). After performing simulations on MLP and PPN, indicators of specificity and sensitivity were 0.88 and 1 in the MLP test and 0.94 and 1 in PPN, respectively. PPN was trained in one round (the advantage of PPN), whereas MLP was trained in 1000 rounds. The comparison of the two networks showed that PPN has the potential to improve the accuracy and speed of classifying patients with coronary artery diseases. The calculated sensitivity and specificity showed that PPN can classify participants into two groups of healthy individuals and patients. Also, prompt diagnosis was made with higher accuracy, compared to similar cases.