Automated prognosis of renal function decline in ADPKD patients using deep learning

• Published in: Zeitschrift für medizinische Physik Vol. 34; no. 2; pp. 330 - 342
• Main Authors: Raj, Anish, Tollens, Fabian, Caroli, Anna, Nörenberg, Dominik, Zöllner, Frank G.
• Format: Journal Article
• Language: English
• Published: Germany Elsevier GmbH 01.05.2024; Elsevier
• Subjects: Adult; Autosomal dominant polycystic kidney disease; Chronic kidney disease; Deep Learning; Disease Progression; Female; Glomerular Filtration Rate; Humans; Image classification; Kidney - diagnostic imaging; Kidney - physiopathology; Magnetic Resonance Imaging - methods; Male; Middle Aged; Original Paper; Polycystic Kidney, Autosomal Dominant - diagnostic imaging; Polycystic Kidney, Autosomal Dominant - physiopathology; Prognosis; Regression; Renal Insufficiency, Chronic - diagnostic imaging; Renal Insufficiency, Chronic - physiopathology; Total kidney volume; Deep learning; Regression; Chronic kidney disease; Total kidney volume; Autosomal dominant polycystic kidney disease; Image classification
• ISSN: 0939-3889; 1876-4436; 1876-4436
• DOI: 10.1016/j.zemedi.2023.08.001

An accurate prognosis of renal function decline in Autosomal Dominant Polycystic Kidney Disease (ADPKD) is crucial for early intervention. Current biomarkers used are height-adjusted total kidney volume (HtTKV), estimated glomerular filtration rate (eGFR), and patient age. However, manually measuring kidney volume is time-consuming and subject to observer variability. Additionally, incorporating automatically generated features from kidney MRI images, along with conventional biomarkers, can enhance prognostic improvement. To address these issues, we developed two deep-learning algorithms. Firstly, an automated kidney volume segmentation model accurately calculates HtTKV. Secondly, we utilize segmented kidney volumes, predicted HtTKV, age, and baseline eGFR to predict chronic kidney disease (CKD) stages >=3A, >=3B, and a 30% decline in eGFR after 8 years from the baseline visit. Our approach combines a convolutional neural network (CNN) and a multi-layer perceptron (MLP). Our study included 135 subjects and the AUC scores obtained were 0.96, 0.96, and 0.95 for CKD stages >=3A, >=3B, and a 30% decline in eGFR, respectively. Furthermore, our algorithm achieved a Pearson correlation coefficient of 0.81 between predicted and measured eGFR decline. We extended our approach to predict distinct CKD stages after eight years with an AUC of 0.97. The proposed approach has the potential to enhance monitoring and facilitate prognosis in ADPKD patients, even in the early disease stages.