Automatic detection of kidneys on abdominal CT images using Aggregate Channel Features

Accurate detection of kidney regions in abdominal CT images makes it easier to detect formations such as cysts, lesions, and stones in the kidneys. In this study, the Aggregate Channel Features (ACF) algorithm, which is a machine learning method, is used for automatic detection of the kidneys. Negative samples are automatically taken from the images during the learning process. The ACF obtained are formed alternately and repeatedly for N steps using the AdaBoost classifier. At each step negative samples are removed and collected with the previous ones. The confusion matrix and k-fold cross-correlation methods are used to test the performance of the study. The data set fragmented according to k-fold is trained according to the location information of the labeled objects using the ACF. Recall, precision, and F1 scores gleaned from the confusion matrix are used in performance analysis. The results show that the proposed method can successfully detect kidney regions.