Seyed Pouya Musavi Ghasemi - Seraj Higher Education Institute-Tabriz
Faramarz Ariyani Shirvanehdeh - Seraj Higher Education Institute-Tabriz
Naser Nasirzadeh Azizkandi - Seraj Higher Education Institute-Tabriz

Chronic kidney disease is a progressive disease associated with a high risk of cardiovascular disease, high mortality and health care costs. Therefore, early diagnosis of the disease is very important to control its consequences. Kidney stones are hard sediments that are often formed due to the increase in the concentration of solutes and salts in the urine. Medical imaging for segmentation of kidney stones is one of the important research topics in recent years. To diagnose kidney stones, the noise of the images and the contrast should be reduced so that the classification and division of the kidney can be done easily. First, we used image pre-processing techniques including noise removal, smoothing, sharpening, and contrast enhacement, then the features were extracted using the gray level co-occurrence matrix, after that the optimal features were selected. In this research, our purpose was to use artificial neural networks along with the Bayesian regularization training algorithm and the K-nearest neighbors algorithm for data classification, where the selected features were considered as input and labels as targets and Finally, according to the results obtained from both methods, the artificial neural networks had a better performance with a ۹۹/۰۷% than K-nearest neighbors algorithm with a ۹۰%.

Chronic kidney disease, kidney stones, artificial neural networks, k-nearest neighbors, CT scan image, classification