Renal Function Evaluation Survey: Artificial Intelligence for the Next Decade?

Abstract

Accurate assessment of renal microstructure and function remains a key point for the prediction and diagnosis of chronic kidney disease (CKD). Applications of novel medical imaging techniques offer a non-invasive and safer tool for analyzing CKD as it allows health care providers to identify morphological, functional and molecular information that detects changes in renal tissue properties and functionalities. Recently, the ability of artificial intelligence to address information retrieval and other critical issues in big medical data analytics has led to a great interest in CKD diagnosis. Besides qualitative analysis of renal medical imaging, texture analysis combined with machine learning has emerged as a promising technique to quantify renal tissue heterogeneity, thus providing a complementary tool for renal function decline prediction. Most importantly, deep learning holds the potential to be a novel approach for renal dysfunction monitoring. This paper proposes a survey focusing on the most recent approaches for using texture analysis and machine learning techniques that can be integrated in clinical research in order to improve renal dysfunction diagnosis and prognosis.