Stage diagnosis for Chronic Kidney Disease based on ultrasonography

Abstract

A large portion of people suffer from Chronic Kidney Disease (CKD) in the world. Unfortunately, some of them don't know they have been contracted CKD until they need dialysis treatment at the end stage. Diagnosis through non-invasive ultrasonic imaging techniques become important clinical approaches for detecting CKD, and high potential or at-risk CKD patients could avoid being infected via blood test and/or reduce chances of abrupt deterioration in renal function by taking iodinated contrast medium. This research established a detection system based on computer vision and machine learning techniques for facilitating diagnosis of CKD and different stages of CKD. Novel features and support vector machine were applied for rapid detection. In this study, several evaluations on different clustered groups were performed and compared according to estimated glomerular filtration rates (GFR). In addition, the proposed system required 0.016 seconds in average for feature extraction and classification for each testing case. The results showed that the system could produce consistent diagnosis based on noninvasive ultrasonographic approaches and which could be considered as the most proper clinical diagnosis and medical treatment for CKD patients.